Official Gradio demo for SeedVR2: One-Step Video Restoration via Diffusion Adversarial Post-Training.
🔥 SeedVR2 is a one-step image and video restoration algorithm for real-world and AIGC content.

---

If you find SeedVR helpful, please ⭐ the GitHub repository:

Notice

Uploaded videos are processed automatically from their detected source length. Each processed download keeps the matching portion of the original audio. The Space offers exact 720p and 1080p exports, while keeping the model on a compatible internal resolution and resizing only at the end.

How it works

Upload an image or a video, choose 720p or 1080p, then click Process next part. The upload is stored only while its resumable job is active. Videos are restored in memory-safe chunks of up to 121 frames. Adjacent chunks share a small frame overlap to reduce visible transitions, then each completed part is joined losslessly before its matching original audio is added back. The progress display shows the current part and chunk, for example Part 3/150 - Chunk 37/150.

ZeroGPU

This Space runs on Hugging Face ZeroGPU. The default five-second part size fits short GPU reservations, so long videos can be continued with repeated clicks without uploading again. Five-minute parts are available for dedicated GPU hardware, but may exceed a ZeroGPU allocation. Use Discard job and parts to remove an unfinished upload and all files created for it.

Limitations

May fail on heavy degradations or small-motion AIGC clips, causing oversharpening or poor restoration.

Citation

        @article{wang2025seedvr2,
            title={SeedVR2: One-Step Video Restoration via Diffusion Adversarial Post-Training},
            author={Wang, Jianyi and Lin, Shanchuan and Lin, Zhijie and Ren, Yuxi and Wei, Meng and Yue, Zongsheng and Zhou, Shangchen and Chen, Hao and Zhao, Yang and Yang, Ceyuan and Xiao, Xuefeng and Loy, Chen Change and Jiang, Lu},
            booktitle={arXiv preprint arXiv:2506.05301},
            year={2025}
        }

        @inproceedings{wang2025seedvr,
            title={SeedVR: Seeding Infinity in Diffusion Transformer Towards Generic Video Restoration},
            author={Wang, Jianyi and Lin, Zhijie and Wei, Meng and Zhao, Yang and Yang, Ceyuan and Loy, Chen Change and Jiang, Lu},
            booktitle={CVPR},
            year={2025}
        }
        

License

Licensed under the Apache 2.0 License.