Image-to-Video
Diffusion Single File
text-to-video
image-text-to-video
video-to-video
text-to-audio-video
image-to-audio-video
image-text-to-audio-video
video-to-audio-video
audio-to-audio-video
audio-video-generation
multimodal
synchronized-audio-video
reference-to-audio-video
Instructions to use rzgar/FastVideo-FastH3-Motion-Enhanced-Comfy with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusion Single File
How to use rzgar/FastVideo-FastH3-Motion-Enhanced-Comfy with Diffusion Single File:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
A ready-to-use FastVideo's FastH3 with motion enhancement baked in.
| File | Link |
|---|---|
| fastvideo_fasth3_8step_v2_MoEn_pruned_int8_convrot.safetensors | Download |
What's different from the base
- more accurate male and female anatomy
- a better understanding of NSFW-related motion
Everything else is the FastH3 8-step model you already know.
Note
Most people don't run a base model on its own. They run their own stack of concept and style LoRAs on top of it.
So instead of pushing the baked-in enhancement all the way to its sweet spot, I deliberately set it to sweet spot − 25%. That leaves headroom for your LoRAs - they still get room to define the look, style and character without fighting the base, while the anatomy and motion enhancment is already there.
That means:
- it works out of the box with no LoRAs at all
- and it stacks cleanly with your own concept/style LoRAs instead of competing with them
- Downloads last month
- 123