| --- |
| license: apache-2.0 |
| base_model: |
| - Wan-AI/Wan2.2-I2V-A14B |
| pipeline_tag: video-to-video |
| --- |
| |
| # VGC: Learning to Correct Geometry in Generated Videos |
|
|
| **NeurIPS 2026** | [Code](https://github.com/weijielyu/VGC) | [Project Page](https://www.wlyu.me/VGC/) |
|
|
| Video Geometry Corrector (VGC) is a post-hoc video-to-video model that corrects geometry distortions in generated videos (warped structures, inconsistent parallax, drifting scene layouts) without modifying the source generator. |
|
|
| For usage instructions, see the [official code repository](https://github.com/weijielyu/VGC). |
|
|
| ## Contents |
|
|
| - **wan22**: Release version of the VGC Wan2.2-I2V-A14B MoE model (720p, no optical flow), fine-tuned from Wan2.2-I2V-A14B |
| - `release_no_flow_high_noise.pth`: high-noise expert |
| - `release_no_flow_low_noise.pth`: low-noise expert |
|
|
| ## Citation |
|
|
| ```bibtex |
| @inproceedings{lyu2026vgc, |
| title={Learning to Correct Geometry in Generated Videos}, |
| author={Lyu, Weijie and Shen, Tiancheng and Li, Xiangtai and Wang, Yujing and Wang, Xiaoyuan and Yuan, Haobo and Zhao, Yizhou and Yang, Ming-Hsuan}, |
| booktitle={Advances in Neural Information Processing Systems (NeurIPS)}, |
| year={2026} |
| } |
| ``` |
|
|
