These are unofficial fp16 versions of
- https://huggingface.co/stabilityai/stable-video-diffusion-img2vid
- https://huggingface.co/stabilityai/stable-video-diffusion-img2vid-xt
They don't seem to reduce VRAM usage, but can save you data & disk.
I couldn't see any difference in generated results compared to the full models (in lowram mode).
Follow me for AI tips & tricks and more: