Vidu S

Introduction

Vidu S2

Vidu S2 extends real-time video generation beyond talking-head digital characters to high-resolution interactive avatars, live video editing, and immersive spatial video. It includes Vidu S2-Avatar for controllable character generation and Vidu S2-Editing for transforming incoming video streams.

Key breakthroughs:

1. 720p real-time interactive avatars
   • Vidu S2-Avatar generates 720p video at 25–42 FPS, follows a wider range of instructions—including large body motions such as dancing—and accepts new reference images at any moment during a stream.
2. Stable long-horizon generation with Self-Replay Forcing
   • Self-Replay Forcing (SRF) replays re-noised, self-generated trajectories in a gradient-enabled causal pass, helping prevent errors from accumulating across streaming segments.
3. Real-time editing of incoming video
   • Vidu S2-Editing supports style transfer, virtual try-on, character replacement, and background replacement from text instructions and optional reference images while preserving the source motion.
4. Real-time spatial video for immersive displays
   • Generated or edited streams can be converted into synchronized left- and right-eye views, while stereoscopic inputs can be edited jointly for streaming to VR headsets.
5. Efficient inference on low-cost GPUs
   • An optimized serving stack combines TurboDiffusion and TurboServe, using efficient attention, low-bit GEMM, kernel and launch optimizations, and multi-GPU pipelining for real-time inference.

Vidu S1

Vidu S1 is a real-time interactive video generation model for voice-controlled digital characters. Users can guide generated video content at any moment through spoken instructions, enabling live interaction.

Key breakthroughs:

1. Real-time speech control over video content
   • Users can directly instruct digital characters to perform actions.
2. Infinite-length real-time interactive generation
   • Vidu S1 generates 540p video at up to 42 FPS and can run on consumer GPUs.
3. Custom character images and voice tones
   • Vidu S1 supports real people, anime-style characters, pets, and other personalized avatars.

Quick Links

Vidu S2

• Try Vidu S2: https://www.vidu.com/vidu-stream
• Vidu S2 Paper: https://arxiv.org/abs/2609.11638

Vidu S1

• Vidu S1 Paper: https://arxiv.org/abs/2607.03118

Docs

English

• User Guide: shengshu.feishu.cn/wiki/X7ZLwyLUzi461LkAlNpcTxs1nsy
• API Documentation: platform.vidu.com/vidu-stream/doc
• Vidu S2-Avatar Quickstart: platform.vidu.com/vidu-stream/doc/s2-avatar/realtime/quick-start
• Vidu S2-Editing Quickstart: platform.vidu.com/vidu-stream/doc/s2-editing/quick-start

Chinese

• User Guide: shengshu.feishu.cn/wiki/Fz8ywkzEwil84LkbCOKcxotMnvb
• API Documentation: platform.vidu.cn/vidu-stream/doc
• Vidu S2-Avatar Quickstart: platform.vidu.cn/vidu-stream/doc/s2-avatar/realtime/quick-start
• Vidu S2-Editing Quickstart: platform.vidu.cn/vidu-stream/doc/s2-editing/quick-start

Updates

• [2026-09]: Vidu S2 is now available to try at vidu.com/vidu-stream
• [2026-07]: Vidu S1 is now available.

Citation

If you find Vidu S useful for your research, please cite:

@article{zhang2026vidus2,
  title={Vidu S2: Real-Time Interactive, Editable, and Spatial Video Generation},
  author={Zhang, Jintao and Jiang, Kai and Chen, Jintao and Wang, Xu and Liu, Deyuan and Li, Jungang and Chen, Dechuang and Lin, Ming and Zhou, Jingjiang and Jin, Haopeng and others},
  journal={arXiv preprint arXiv:2609.11638},
  year={2026}
}

@article{zhang2026vidus1,
  title={Vidu S1: A Real-Time Interactive Video Generation Model},
  author={Zhang, Jintao and Jiang, Kai and Chen, Jintao and Wang, Xu and Luo, Yang and Wang, Yuji and Chen, Dechuang and Li, Jungang and Ye, Chengyang and Chen, Marco and others},
  journal={arXiv preprint arXiv:2607.03118},
  year={2026}
}