Wan 2.2 · image → video
Make a still image move.
Drop a photo, write a short prompt, get a few seconds of video. Runs on a public Hugging Face Space, so it's free, queued, and surprisingly capable.
Describe motion and camera, not just what's in the frame. The model already sees the image.
Advanced options
Capped at 4.5s. The free upstream Space hits its wall-clock limit around 5s and returns nothing.
More steps = better quality but slower.
Add an image to start.
Generated video will appear here.
Warming up the GPU...
First call cold-starts the Space, which can take a minute or two.
This session