Upload a clip, describe what should change, and Genjutsu AI rewrites it — a new outfit, a new character, new weather, a new world — while aiming to preserve the camera movement and timing of your original footage. Results can change details you wanted to keep.
Preview the setup before paying. Sign in and use credits only when you generate.
Neon dancer → 3D arctic foxSee all examples
Add a reference image of the new character, then say what should stay.
Describe the change
Effect
Reference images (1–8)
Photos of what goes in: the new outfit, character, product or car.
Choose your AI model
Resolution
Pro runs Higgsfield Genjutsu through its official API. It is the option for this specific model; a better result is not guaranteed.
Trial pack · $4.99· 300 creditsCredits are based on clip length, rounded up to whole seconds. AI output can change faces, hands, backgrounds or framing; a completed result is charged even if you dislike it.
Signing in or buying credits temporarily saves this setup on this device. Your files are not uploaded until you generate.
No clip? Start from a scene
Tap a scene to load its clip, reference and prompt into the editor, then swap in your own photo.
Ideas to try
Real renders from this editor, the original clip next to its AI-edited result and playing in step. Pro examples use Higgsfield Genjutsu; Standard examples use Kling O3 or FLUX.3 Edit. Press Recreate to load the same clip, references and prompt into the editor.
Replace a person, product or outfit with the one in your reference image. The original movement, camera and lighting guide the edit, but details can change.
“Replace the dancer with the character from the reference image. Keep every arm and body movement, the timing, the camera movement and the neon-lit background exactly as in the original video.”
Source footage: Pexels · References: AI-generated
Keep your clip's movement and give it a new performer or setting. Your reference images decide who moves and where.
“The robot from the reference image performs the exact same dance on the same night basketball court. Keep the camera, framing, timing and the green court lights.”
Source footage: Mixkit · References: AI-generated
Re-render the whole clip in a new look — knitted yarn, hand-painted anime, clay and more — while aiming to retain the action and camera movement.
“Keep both skaters' moves, the palm-lined boardwalk and the camera.”
Source footage: Pexels
More before-and-after renders from Pro and Standard, each labelled with its actual model, prompt and references. Recreate any of them as is, or swap in your own clip or character.
“Replace the model on the runway with the corgi from the reference image, walking upright toward the camera with the same steps and pace. Keep the stage, the spotlights, the camera and the timing.”
“Replace the man with the orange cat from the reference image, walking upright on its hind legs toward the camera with the same stride. Keep the teal corridor, the yellow backlight, the camera and the timing.”
“Replace the white car with the mint-green bubble car from the reference image, driving the same line around the bend with its headlights on. Keep the road, the sunset backlight, the trees and the camera move.”
“Replace the smartwatch and its red strap with the gold diamond watch and gold bracelet from the reference image. Keep the hand, the finger tap, the lighting and the camera.”
“Replace the man's denim shirt with the silver puffer jacket from the reference image, worn open over his white T-shirt. Keep his face, the phone, the street and the camera.”
“The character from the first reference image performs the exact same dance on the rainy neon street from the second reference image. Keep the original camera movement, framing and timing.”
“Keep the slow camera pan.”
Make your own version with a real Genjutsu render.
300 credits: one 5-second Pro render or two Standard Swaps. One-time, nothing renews.
Try it for $4.99Choose Standard or Pro to see the clip requirements. Pro accepts MP4 clips from 4 to 30 seconds, up to 200 MB; Standard takes up to 15 seconds, with limits that depend on the effect. Start with a steady shot and a clearly visible subject.
Say what to swap and what to keep: “replace the sedan with a red vintage convertible, keep the camera pan and the people walking by.”
Genjutsu AI re-renders every frame of the clip and hands back an MP4 at 480p or 720p, ready to post or drop into your edit.
Good Genjutsu AI prompts name one change and say what must stay. Load any of these into the editor and swap in your own details.
Browse 40+ prompts by categoryChange a hoodie into a leather jacket or a suit into a kimono, with the fabric moving the way the original did.
“Change his grey hoodie to a deep red windbreaker. Keep the face, hair and background untouched.”
Put a different person, creature or robot into the same performance — guided by the original pose, timing and camera movement.
“Turn the dancer into a chrome humanoid robot with the same choreography and the same lighting.”
Turn a sunny street into a rainy night, add snowfall, or move a scene to golden hour without a reshoot.
“Turn this sunny street into a rainy night: wet asphalt, neon reflections, light drizzle, cool blue grade.”
Move a skater from a parking lot to a neon Tokyo plaza, or a talking head from an office to a seaside terrace.
“Move the skateboarder to a Tokyo plaza at dusk with neon signs; keep his tricks and the tracking shot.”
Replace a parked sedan with a vintage convertible, a mug with a product can, a dog with a fox.
“Replace the grey sedan with a cherry-red 1960s convertible with chrome trim, top down; keep the camera pan.”
Restyle a whole clip as hand-painted anime, claymation or a comic look while keeping the original motion.
“Restyle the whole clip as a hand-painted anime scene; keep the original motion and timing.”
Choose Pro when you want the Higgsfield Genjutsu model. Standard uses Kling O3 or FLUX.3 Edit at a lower credit cost. Both offer video editing, but quality varies by task; Pro is not guaranteed to produce a better result.
| Standard | Pro (Genjutsu) | |
|---|---|---|
| Model | Kling O3 (Swap, Motion) · FLUX.3 Edit (Restyle) | Higgsfield Genjutsu (official API) |
| Restyle looks | 8 built-in styles | 20 Genjutsu presets |
| Reference images | Up to 4 (none for Restyle) | Up to 8 |
| Best for | Cheaper drafts and everyday edits | Using the Higgsfield Genjutsu model |
| Source clip | Swap & Motion need 720p or larger | 480p or larger |
| Resolution | Up to 720p | 480p or 720p |
| Clip length | 4 – 15 seconds | 4 – 30 seconds |
| Credit cost | 10 – 25 credits / second | 50 – 120 credits / second |
Why Genjutsu AI Video
The simplest way to use the Genjutsu model
Upload a clip, write one sentence, render. No timeline, no keyframes, no app to install. Not sure what to write? Improve prompt writes it for you.
Know the cost first
See the exact credits before you render. Failed renders are refunded automatically.
Your videos stay private
Your renders are never posted to a public gallery or page.
A monthly plan gives the most credits for your money and can be canceled anytime; one-time packs never renew. You see the exact cost on the button before every render, and failed renders are refunded automatically.
Try real Genjutsu renders
$4.99one-time
Lite
Save 22% vs one-time packs
800 credits cost $12.69 as a pack. Here: $9.90 a month (1.24¢ per credit).
A few videos every month
Plus
Save 28% vs one-time packs
1,750 credits cost $27.77 as a pack. Here: $19.90 a month (1.14¢ per credit).
For regular creators
Max
Save 32% vs one-time packs
3,700 credits cost $58.71 as a pack. Here: $39.90 a month (1.08¢ per credit).
For client and ad work
Genjutsu AI is video-to-video editing: you give it footage you already have, and it changes what is in the frame while using the original shot as a guide. The original footage guides the performance, camera movement and timing. You can ask to change clothes, characters, objects, location, weather or visual style, but the model can also alter faces, lighting, framing or other details you wanted to preserve.
The name comes from Higgsfield AI Genjutsu — Higgsfield Genjutsu for short — the video-to-video model Higgsfield released in September 2026, which in turn borrows the Naruto term for an illusion cast over what someone sees. The footage is real; what you see in it is not. “Genjutsu AI” has since become the way people search for this kind of editing.
GenjutsuAIVideo.com is an independent Genjutsu AI editor built around that workflow. Pro mode runs Higgsfield AI Genjutsu itself through its official API. Standard mode runs lighter models — Kling O3 for Swap and Motion transfer, FLUX.3 Edit for Restyle — at a half to a sixth of the Pro price, so you can choose a model based on your task and budget.
Think of every clip as two layers. The first is how it was shot: camera motion and framing, the actor’s movement and expressions, lip-sync, pacing and edit points, where the light falls. Genjutsu AI uses that layer as a guide; it does not guarantee that every detail survives. The second layer is what the shot contains: wardrobe, faces and characters, props and vehicles, the location, the weather, the time of day, the overall look. That is the layer your prompt rewrites.
Because the motion is borrowed rather than invented, results feel like the same take filmed in a different world — not like a new, slightly random generation. That is the main reason creators reach for Genjutsu AI instead of plain text-to-video.
Text-to-video models invent motion from nothing. You describe a shot and get one possible version of it; if the actor’s timing or the camera move is wrong, you roll the dice again. Video-to-video editing starts from a take you already like, so the hardest parts — performance and camera work — are already solved.
That makes Genjutsu AI the better fit when you have a great take with the wrong outfit, a product shoot you want in five seasonal settings, a short you want to restyle as anime, or a continuity error you need to fix without a reshoot.
Genjutsu AI is a video-to-video editor. You upload a clip and describe a change — a new outfit, character, location, weather or style — and it re-renders the clip with that change while aiming to retain the original camera motion and timing. Faces, backgrounds, lighting and framing can still change.
There is no free tier: every render runs on a paid video model. The cheapest way to try it is the $4.99 trial pack — 300 credits, enough for one 5-second Pro (Genjutsu) render at 480p or two 5-second Standard Swaps. Packs are one-time and never renew, and failed renders are refunded.
No. GenjutsuAIVideo.com is an independent tool and is not affiliated with or endorsed by Higgsfield. Pro mode calls the Higgsfield Genjutsu model through its official API, so you get the same model without a subscription: pay only for the seconds you render, see the cost before you generate, and get credits back automatically if a render fails.
Higgsfield AI Genjutsu is the video-to-video model Higgsfield released in September 2026. “Genjutsu AI” is what most people search for when they mean that kind of editing, and it is the name of this editor. Here, Pro mode runs Higgsfield AI Genjutsu through its official API; Standard mode runs lighter models — Kling O3 and FLUX.3 Edit — for cheaper drafts. Every mode is labelled with the model it uses.
No credits are deducted for requests rejected before submission. Submitted renders that fail, are blocked by provider moderation or are confirmed canceled return their credits automatically. On status checks, we try to cancel jobs queued over 30 minutes; a return follows only when cancellation succeeds. Jobs still unfinished after two hours are marked failed and refunded. A successful render is charged even if you dislike the result. Returning render credits is separate from refunding an unused credit pack.
Usually a few minutes. Longer clips and 720p output take longer. You can leave the page while it renders — finished videos are saved to your account.
No. It works on ordinary footage. It does best when the subject you want to change is clearly visible and not hidden behind other objects for most of the clip.
Yes. Use Swap for one character or object, or Motion transfer to drive a new character with the movement from your clip. Describe the new character in detail and say what should stay the same.
Clips of 4 to 30 seconds, at least 480p (854×480), work as they are: an MP4 up to 200 MB (Standard: up to 15 seconds; Swap and Motion transfer need 720p or larger, Restyle files up to 50 MB). Longer clips, larger files and iPhone .mov videos can be trimmed and converted to MP4 right in the editor — it runs in your browser, nothing is uploaded until you press Generate. Renders come out at 480p or 720p, and you pay per second of your clip.
Your clip and reference images go straight to the AI model provider’s storage to run the render; we don’t keep our own copy. Finished videos are saved to My videos in your account for at least 90 days — download the ones you want to keep, as older files may be removed. Renders are private and never published. To delete your videos or your account sooner, email support@genjutsuaivideo.com from your account address.
Generative editing can change details even when you ask to keep them. Faces, hands, occluded subjects and fast movement are especially difficult. Start with a steady clip and one clear subject, name one change, and state what should stay. These steps can help, but do not guarantee an unchanged face or background.
Sexual content, anything involving minors, deepfakes or impersonation of real people, and footage you don’t have the rights to. Prompts rejected before submission are not charged. If provider moderation blocks a submitted render, its credits are returned automatically. Repeated prohibited use can lead to account suspension. Public figures and copyrighted characters are often blocked too, even when you have permission.
You can use the videos you generate from footage you have the rights to. Do not upload footage of people without their consent or content you are not allowed to modify — see our Terms of Service.
Upload a clip, write one sentence, and let Genjutsu AI show you the same take in a different world.
