README.md

---
license: other
license_name: stabilityai-community-license
license_link: https://huggingface.co/stabilityai/stable-video-diffusion-img2vid/blob/main/LICENSE.md
library_name: diffusers
pipeline_tag: image-to-image
---

# VideoMaMa: Mask-Guided Video Matting via Generative Prior

[**Sangbeom Lim**](https://sites.google.com/view/sangbeomlim/home) · [**Seoung Wug Oh**](https://sites.google.com/view/seoungwugoh) · [**Jiahui Huang**](https://gabriel-huang.github.io/) · [**Heeji Yoon**](https://yoon-heez.github.io/) · [**Seungryong Kim**](https://cvlab.kaist.ac.kr/members/faculty) · [**Joon-Young Lee**](https://joonyoung-cv.github.io)

[[Paper](https://huggingface.co/papers/2601.14255)] [[Project Page](https://cvlab-kaist.github.io/VideoMaMa/)] [[GitHub](https://github.com/cvlab-kaist/VideoMaMa)] [[Gradio Demo](https://huggingface.co/spaces/SammyLim/VideoMaMa)]

VideoMaMa (Video Mask-to-Matte Model) is a framework that converts coarse segmentation masks into pixel-accurate alpha mattes by leveraging pretrained video diffusion models. It demonstrates strong zero-shot generalization to real-world footage, even though it is trained solely on synthetic data.

## Inference

To use VideoMaMa for inference, you can use the script provided in the [official repository](https://github.com/cvlab-kaist/VideoMaMa):

```bash
python inference_onestep_folder.py \
    --base_model_path "stabilityai/stable-video-diffusion-img2vid-xt" \
    --unet_checkpoint_path "SammyLim/VideoMaMa" \
    --image_root_path "/path/to/your/images" \
    --mask_root_path "/path/to/your/masks" \
    --output_dir "./output" \
    --keep_aspect_ratio 
```

## License

The VideoMaMa model checkpoints (specifically `unet/*` and `dino_projection_mlp.pth`) are subject to the **Stability AI Community License**. By using this model, you agree to the terms outlined in the [license agreement](https://huggingface.co/stabilityai/stable-video-diffusion-img2vid/blob/main/LICENSE.md).

## Citation

```bibtex
@article{lim2026videomama,
  title={VideoMaMa: Mask-Guided Video Matting via Generative Prior},
  author={Lim, Sangbeom and Oh, Seoung Wug and Huang, Jiahui and Yoon, Heeji and Kim, Seungryong and Lee, Joon-Young},
  journal={arXiv preprint arXiv:2601.14255},
  year={2026}
}
```