FramePack-F1

Video diffusion, but feels like image diffusion

FramePack F1 - a FramePack model that only predicts future frames from history frames

beta FramePack Fill 🖋️- draw a mask over the input image to inpaint the video output

adapted from the officical code repo FramePack by lllyasviel and FramePack Studio 🙌🏻

Quick List
1 5

Faster speed, but often makes hands and fingers slightly worse.

1 100
1 32
6 128
0 100
Share your results and find ideas at the FramePack Twitter (X) thread