Genjutsu AI Video Editor: change an existing clip with a reference and prompt
GenjutsuAIVideo.com is an independent video-to-video editor. You provide existing footage, describe a change and optionally add reference images. The service generates an edited clip using the source movement and camera work as guidance, while warning that details you wanted to preserve can change.
Choose a model and one clear change
The Genjutsu AI Video Editor homepage offers Swap, Motion transfer and Restyle effects. A swap might replace a product or outfit; motion transfer applies the source performance to a new character or setting. Restyle changes the visual treatment of the whole clip.
The publisher says Pro runs Higgsfield Genjutsu through its official API. Standard uses Kling O3 for Swap and Motion transfer, and FLUX.3 Edit for Restyle. The site is not affiliated with or endorsed by Higgsfield, and the Pro label does not guarantee a better result.
Describe the change and the elements that should remain. The homepage recommends a steady shot and a visible subject. Review faces, hands and background details afterward, since generative edits can alter them even when the prompt asks otherwise.
Check the input and output limits
Pro accepts MP4 clips from four to thirty seconds, up to 200 MB, at the stated minimum resolution. Standard allows up to fifteen seconds with effect-specific limits; Swap and Motion transfer need at least 720p source footage. The publisher also describes browser tools for trimming and converting larger or differently formatted inputs.
Output is an MP4 at 480p or 720p. Those choices matter if the result must fit into a higher-resolution production: inspect the final clip before assuming it can replace the source shot without additional finishing.
The homepage's examples include original clips beside generated results and label their models. This overview follows those published descriptions; we have not watched every render, generated a clip or verified identity preservation.
Understand credits, retention and consent
The service uses one-time credit packs rather than a subscription. The captured entry offer is $2.99 for 300 credits, with the exact generation cost shown before rendering. Credits depend on clip length rounded up to whole seconds, model and effect.
Failed renders return credits under the publisher's stated rules. A successful result is charged even if you dislike it. That distinction makes a small representative test more useful than assuming every paid attempt will produce an acceptable edit.
The publisher says source clips and references go to the model provider's storage, while finished results stay in the account for at least 90 days. Download files you need to keep and review the terms before uploading client material. Use footage you have permission to modify and obtain consent for identifiable people. The service prohibits impersonation and other restricted content; a reference image does not grant rights to its subject.