Genjutsu-style motion tricks, built on open Hugging Face models. Free forever.
Depth Anonymizer — upload any video and get back a depth-map version.
Every movement is preserved, but faces and identities are erased. It runs
entirely in your browser (Depth Anything via Transformers.js) — your
video never leaves your device.
1. Pick a video
Choose a video to begin. Short clips (under 12 seconds) work best.
OriginalDepth map
Want the full motion transfer?
Turning a character photo into a moving video needs a GPU. The free way to
try it: download
motion-studio-colab.ipynb,
upload it to Google Colab
(File → Upload notebook), and run it on a free GPU. The notebook fetches
the Motion Studio app from this Space and launches it with a share link.