Changelog-Style Notes: What Motion Transfer Ships Todayannouncement

What is available today

Genjutsu AI performs video-to-video motion transfer only, not text-to-video. A real source clip supplies the movement; reference images define the character or subject performing it.

Supported inputs and outputs

• Source clip: 4–30 seconds, in MP4 or MOV format.
• Reference images: up to 20 per generation.
• Output: 480p, 720p, or 1080p.
• Credit rates: 3, 7, or 11 credits per second at those respective resolutions.

Credits and automation

One-time credit packs are prepaid, with no subscription. Credits never expire, and failed generations are refunded. A REST API is available for automation.

Videos made before a first purchase are public in Community, so use disposable footage until you buy a pack.

Genjutsu AI launched on 2026-10-09.

A practical first render

Start with a short clip featuring a single subject and readable motion. Choose consistent reference images and test at 480p before increasing resolution. Review difficult motion frame by frame to identify input problems early.

— Genjutsu AI