vita-epfl/Stable-Video-Infinity
Research project presenting a video diffusion transformer that generates extended videos via iterative error recycling across overlapping windows.
Not currently ranked — collecting fresh signals.
star history
This repository implements Stable Video Infinity, a video generation method using diffusion transformers to produce arbitrarily long videos. The approach cycles overlapping generated segments back into the denoising process to correct accumulated errors, enabling coherent multi-segment video generation. It supports applications including audio-driven talking faces and dance generation, and provides model weights on HuggingFace.
Frequently asked
- What is vita-epfl/Stable-Video-Infinity?
- Research project presenting a video diffusion transformer that generates extended videos via iterative error recycling across overlapping windows.
- Is Stable-Video-Infinity open source?
- Yes — vita-epfl/Stable-Video-Infinity is open source, released under the MIT license.
- What language is Stable-Video-Infinity written in?
- vita-epfl/Stable-Video-Infinity is primarily written in Python.
- How popular is Stable-Video-Infinity?
- vita-epfl/Stable-Video-Infinity has 2.5k stars on GitHub.
- Where can I find Stable-Video-Infinity?
- vita-epfl/Stable-Video-Infinity is on GitHub at https://github.com/vita-epfl/Stable-Video-Infinity.