Upload a short clip and one photo. Get the same moves and camera with a new performer, or swap one person, object or background.
Powered by leading AI video models
Genjutsu AI is an AI video editor for footage that already exists. You bring a real performance — a dance from a trend, a parkour run, a product demo, a talking-head take — and it makes a new version of that exact shot. In ninja fiction, a genjutsu is an illusion: the viewer sees something different while the real movement underneath never changes. That is the whole idea here.
There are two ways to use it. Motion Transfer lifts the movement and camera path out of your clip and hands them to the character in your reference image, so the person and the world are new but the performance is yours. Object Swap keeps the shot you filmed and replaces one target — a person, an object or the background. Clips run from 4 to 30 seconds for Motion Transfer and 4 to 15 seconds for Object Swap, which suits single shots, short-form posts, ad variants and music-video moments.
Film a move once and hand it to someone new. Motion Transfer carries body movement, timing and camera path from your clip into a fresh character and setting.
Object Swap is for shots that already work. Pick Person, Object or Background, upload the replacement, and the rest of the frame stays as filmed.
Run the same source clip with a different presenter, product or location for each market, audience or A/B test.
Viral dance and performance edits are built from one shared source clip. Use the same clip with your own character and post your version while the trend is still moving.
Test an idea at a lower resolution, then render the keeper at full quality. Motion Transfer and Object Swap both offer 720p and 1080p.
Three steps. No timeline, no masks, no keyframes.
Upload an MP4 or MOV between 4 and 30 seconds (up to 15 seconds for Object Swap). This clip decides the movement, timing and camera path of the result.
For Motion Transfer, add a JPG or PNG of the new character and describe the scene. For Object Swap, add the replacement and choose Person, Object or Background.
The Generate button shows the credit cost before you start. Download the finished video from your list, or run it again with another reference.
Your clip supplies the movement and the camera. The reference image supplies who is moving and where.
Choose a person, an object or the background to replace. Everything else stays exactly as you filmed it.
Characters, outfits, props and locations all start from one clear reference image.
Each example below was made with Motion Transfer or Object Swap. Open one to load its setup into the generator.
Buy credits once and use them whenever you need them. No subscription. Each generation shows its credit cost before you start.
Credits never renew automatically
≈ $5.9 per video
Failed renders never cost credits · Credits never expire
Credits never renew automatically
≈ $5.0 per video
Failed renders never cost credits · Credits never expire
Credits never renew automatically
≈ $4.7 per video
Failed renders never cost credits · Credits never expire
Pack credits stay in your account until you use them.
Accepted payment methods
Secure checkout powered by Stripe
Still stuck? Reach us through the contact page.
It's an online AI video editor that works video-to-video. Give it a short clip (4–30 seconds, or 4–15 seconds for Object Swap) and a reference image, and it either moves the clip's performance onto a new character and scene (Motion Transfer) or replaces one person, object or background while the rest of the shot stays put (Object Swap).
No. Genjutsu AI is an independent product with no connection to Higgsfield. It does the same kind of edit — motion transfer and object swap — on footage you upload yourself.
Use Motion Transfer when you want a new performer or a new world but the same movement. Use Object Swap when the shot is already right and only one element — a person, a product or the background — needs to change.
Yes, in Object Swap. Before generating, pick Person, Object or Background as the target. Motion Transfer works differently: it rebuilds the whole scene around the original movement.
Yes. Use Object Swap with Background selected and upload the new setting as your reference. You, your movement and the camera stay as filmed.
Usually 10–20 minutes. Object Swap clips tend to come back faster; Motion Transfer takes longer, especially at 1080p or with a full 30-second clip. Rendering happens in the background, so you can keep browsing and find the finished video in your history. If a render fails, the credits go back to your balance automatically.
The video models refuse some inputs, such as protected footage or restricted subjects. The render fails and your credits come back. Clips you filmed yourself are the most reliable source.
Source video: MP4 or MOV, up to 95 MB, 4–30 seconds for Motion Transfer or 4–15 seconds for Object Swap. Object Swap videos must be at least 700 px wide and tall. Reference image: JPG or PNG, up to 10 MB for Motion Transfer or 20 MB for Object Swap.
One main subject, steady framing, good light and full-body movement that stays in frame. Heavy motion blur, people overlapping and hands vanishing behind clothing are the hardest moments for the model to rebuild.
Credits from packs and the sign-up bonus never expire.
None. There is no timeline, masking or keyframing — you add a clip and a reference, pick a mode, describe the change and generate.
Bring one clip and one reference image. Genjutsu AI keeps the performance and changes what you ask for — and tells you the price before it starts.
