magic-research/magic-animate
A diffusion model that animates a reference human image using pose sequences to produce temporally consistent videos.
Not currently ranked — collecting fresh signals.
star history
MagicAnimate is a diffusion-model-based system for animating still human images into videos using pose guidance. The method ensures temporal consistency across frames by leveraging a diffusion architecture trained on human motion data. The repository includes inference code, a Gradio demo, and pretrained model weights for Stable Diffusion V1.5 integration.
Frequently asked
- What is magic-research/magic-animate?
- A diffusion model that animates a reference human image using pose sequences to produce temporally consistent videos.
- Is magic-animate open source?
- Yes — magic-research/magic-animate is open source, released under the BSD-3-Clause license.
- What language is magic-animate written in?
- magic-research/magic-animate is primarily written in Python.
- How popular is magic-animate?
- magic-research/magic-animate has 10.9k stars on GitHub.
- Where can I find magic-animate?
- magic-research/magic-animate is on GitHub at https://github.com/magic-research/magic-animate.