README.md

metadata

title: LiveEdit
emoji: 🎬
colorFrom: purple
colorTo: blue
sdk: gradio
sdk_version: 5.49.1
app_file: app.py
python_version: '3.10'
short_description: Real-time diffusion-based streaming video editing (Wan2.1)
startup_duration_timeout: 1h
pinned: false

LiveEdit — Towards Real-Time Diffusion-Based Streaming Video Editing

A ZeroGPU demo of LiveEdit (ECCV 2026).

LiveEdit is a diffusion-based framework for streaming video editing. Given a source video and a text editing instruction, it performs causal chunk-by-chunk editing while preserving backgrounds and non-edited regions. It is built on Wan2.1-T2V-1.3B and the Self-Forcing codebase, distilled to a few-step streaming student.

• Project page: https://live-edit.github.io
• Paper: https://arxiv.org/abs/2606.26740
• Checkpoints: https://huggingface.co/cp-cp/LiveEdit

This Space uses the released ar-forcing_002000.pt checkpoint with the default 4-step causal editing configuration. The source video is resampled to 81 frames at 480×832; the output is ~5 seconds at 16 fps.

Citation

@inproceedings{wang2026liveedit,
  title={LiveEdit: Towards Real-Time Diffusion-Based Streaming Video Editing},
  author={Wang, Xinyu and Zhao, Chongbo and Zhan, Fangneng and Ma, Yue},
  booktitle={European Conference on Computer Vision},
  year={2026},
  organization={Springer}
}