PointDiT: Pixel-Space Diffusion for Monocular Geometry Estimation ================================================================= This directory is released under the Apache License 2.0 of the enclosing repository (see the LICENSE file at the repository root). It also contains code derived from the third-party projects listed below, which remain subject to their own licences. The notices reproduced here satisfy those licences' attribution requirements. ----------------------------------------------------------------------- 1. JiT -- https://github.com/LTH14/JiT MIT License, Copyright (c) 2025 Tianhong Li ----------------------------------------------------------------------- This project began as a fork of JiT, which contributed the pixel-space diffusion transformer backbone, the training loop and the command-line interface. Files substantially derived from it: main.py, engine.py, model.py, denoiser.py, util/misc.py, util/lr_sched.py, util/model_util.py. Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. ----------------------------------------------------------------------- 2. MoGe -- https://github.com/microsoft/MoGe MIT License, Copyright (c) Microsoft Corporation ----------------------------------------------------------------------- The evaluation protocol and the affine alignment code are adapted from MoGe: metrics/alignment.py, metrics/compute_metrics.py, and the depth I/O helper `read_depth` in dataloader/eval_depth.py. The pre-processed evaluation datasets are also distributed by the MoGe authors. MIT terms as reproduced in section 1 above, with copyright held by Microsoft Corporation. ----------------------------------------------------------------------- 3. utils3d -- https://github.com/EasternJournalist/utils3d MIT License, Copyright (c) Ruicheng Wang ----------------------------------------------------------------------- The mask-aware resizing and tensor helpers used by the metrics are adapted from utils3d: metrics/masked_resize.py, metrics/utils.py, metrics/helpers.py. MIT terms as reproduced in section 1 above. ----------------------------------------------------------------------- 4. Depth Pro -- https://github.com/apple/ml-depth-pro Copyright (C) 2024 Apple Inc. All Rights Reserved. ----------------------------------------------------------------------- metrics/boundary_metrics.py implements the scale-invariant boundary F1 metric and is vendored from Depth Pro. Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met: 1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer. 2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution. 3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. ----------------------------------------------------------------------- 5. SiT and Lightning-DiT https://github.com/willisma/SiT (MIT), https://github.com/hustvl/LightningDiT ----------------------------------------------------------------------- Transformer design details in model.py follow SiT and Lightning-DiT, as recorded in that file's header. Reached via the JiT fork in section 1. ----------------------------------------------------------------------- Note on DINOv3 ----------------------------------------------------------------------- PointDiT conditions on a frozen DINOv3 encoder. No DINOv3 code or weights are vendored in this directory: the encoder is loaded from a user-supplied local checkout of https://github.com/facebookresearch/dinov3, and its LVD-1689M weights are access-gated by Meta and must be obtained separately under the DINOv3 License Agreement. See README.md for setup.