Recast a dance or trend clip
Keep the choreography and camera of a dance, sketch or trend video and put a different character in it: another person who agreed to it, a mascot, a game character or an anime hero.
Change what is in the shot. Keep the motion.
Genjutsu AI Video Generator
Genjutsu AI is reference-guided video-to-video editing: start from an existing clip, add up to four reference images, describe the change, and replace a character, outfit, product or background, or restyle the whole shot, while the original motion, camera movement and timing stay in place. This is an independent tool powered by Seedance 2.5 and MiniMax H3, not affiliated with Higgsfield.
Start a Genjutsu edit- Upload or use a built-in clip
- 5, 10, 15 or 30 seconds
- 480p–1080p MP4 in your clip’s shape
- Keeps the source audio
What Genjutsu AI can change in your video
Every edit starts from footage you already have. Pick one kind of change; the rest of the shot is kept as close to the source as the model allows.
| Change | What you add | What changes | What stays |
|---|---|---|---|
| Character | Clip + 1–4 images of the new character | The main person or character: face, hair, body, costume | Motion, camera move, timing, background |
| Outfit | Clip + images of the clothes | Clothing and accessories | The person’s face and hair, pose, movement, shot |
| Product | Clip + images of your product | The featured product or object | Hands, camera move, timing, scene |
| Background | Clip + images of the new place | Background and environment | People, their outlines and actions, camera |
| Style | Clip + a described look (images optional) | Rendering style: anime, 3D, clay, film look | Subjects, composition, performance, timing |
How to make a Genjutsu AI video
- 01
Pick the source and the moment
Use a built-in clip or upload an MP4 or MOV up to 5 minutes. Choose 5, 10, 15 or 30 seconds and drag the window to the moment you want.
- 02
Choose the change and add images
Select character, outfit, product, background or style. Add up to four reference images of what should appear; a style change can be described in words alone.
- 03
Describe it in one sentence
Say what changes and which image to use, for example who Image 1 replaces. Short, specific instructions work best.
- 04
Choose model and resolution, then generate
Pick Seedance 2.5 or MiniMax H3 and 480p, 720p or 1080p. The credit cost is shown before you start; download the MP4 when it is ready.
Genjutsu AI specs and credits
Exact limits, so you know before you upload. Credits are confirmed again when you submit, and a failed generation returns them.
| Item | Value |
|---|---|
| Source video | Built-in clip, or your MP4/MOV: up to 50 MB, 5 seconds to 5 minutes, 300–4096 px per side |
| Edited length | 5, 10, 15 or 30 seconds (MiniMax H3: up to 15 seconds), starting at any second |
| Reference images | Up to 4 JPG, PNG or WebP images, 10 MB each; optional for a style change |
| Output | MP4 at 480p, 720p or 1080p (MiniMax H3: 720p or 1080p), in your clip’s shape: 16:9, 4:3, 1:1, 3:4 or 9:16 |
| Audio | The selected section’s original audio is restored; silent clips stay silent |
| Models | Seedance 2.5 video edit (default) or MiniMax H3 reference generation |
| Watermark | Accounts without a paid order get a moving site watermark; the clean original unlocks after any purchase |
| Failures | One charge per video; a failed generation or rejected material returns the credits |
Credits per video
| Model · resolution | 5 秒 | 10 秒 | 15 秒 | 30 秒 |
|---|---|---|---|---|
| Seedance 2.5 · 480p | 39 | 78 | 117 | 233 |
| Seedance 2.5 · 720p | 88 | 175 | 262 | 523 |
| Seedance 2.5 · 1080p | 155 | 310 | 464 | 928 |
| MiniMax H3 · 720p | 39 | 77 | 116 | — |
| MiniMax H3 · 1080p | 62 | 123 | 185 | — |
Credits come from any plan or credit pack and are shared with every tool on this site.
View credit plansWhat stays the same, and where it can drift
Seedance 2.5 edits the selected section itself, so body motion, gestures, cuts, camera movement and framing follow the source most closely. MiniMax H3 generates a new shot that follows your clip as a reference, which allows bigger changes but can loosen exact timing. Either way the result is generated, not pixel-identical to the source.
For cleaner results
Use a steady, well-lit clip with one clear subject, full-body images for character swaps, plain-background product photos, and one change per generation.
Known limitations
- Fingers wrapping a product or crossing in front of a face can merge or change shape for a few frames.
- Long hair and loose clothing may differ around fast spins and head turns.
- Small logos and packaging text are not guaranteed to stay exact.
- With several similar-looking people, say clearly which one changes, for example the dancer on the left.
- A replacement with very different proportions from the original performer can drift more.
Popular Genjutsu AI edits
Swap the product in an ad
Reuse a finished product shot, with the same hand movement and camera path, for a different bottle, box or gadget. Plain-background product photos from two or three angles help.
Try another outfit
Show a garment on the same person in motion without a reshoot, for fashion previews, lookbooks and social posts.
Turn footage into anime or 3D
Restyle a real clip as hand-drawn anime, a 3D render or claymation while the performance and shot stay the same.
Move the action somewhere else
Put the same performance in a neon street, a beach or a studio set, with lighting matched to the new place.
Genjutsu-style editing vs face swap vs text-to-video
| Capability | Genjutsu-style edit | Face swap | Text or image to video |
|---|---|---|---|
| Starts from your existing clip | Yes | Yes | No |
| Keeps the original motion and camera | Yes | Yes | Generates new motion |
| Replaces the whole character, body and costume | Yes | Face only | No |
| Changes outfit, product or background | Yes | No | No |
| Restyles the whole shot | Yes | No | Yes |
Use footage you may edit, on an independent tool
Edit only videos and images you have the rights to use. Anyone recognizable must be an adult who agreed to the edit, and content must stay non-sexual. Label AI edits as AI-made when you share them, and do not use the tool to impersonate or deceive.
Reference images of people and the selected clip go through the provider’s asset review before Seedance 2.5 generation; rejected material stops the job and returns the credits. These checks are not a guarantee of safety or permission, and there is no separate review of the finished video.
Genjutsu AI questions
What is Genjutsu AI?
Genjutsu AI is the name people use for reference-guided video-to-video editing: you keep an existing clip’s motion, camera movement and timing, and change what is in it, such as the character, outfit, product, background or art style. On this page you upload or pick a clip, add reference images, describe the change and download an MP4.
Is this the official Higgsfield Genjutsu?
No. This is an independent tool and is not affiliated with, endorsed by or operated by Higgsfield. It runs Seedance 2.5 video editing and MiniMax H3 through this site’s own pipeline, with its own credits and limits.
Can I replace a person in a video without changing the movement?
Yes. Choose Character, add one to four images of the new character, and say who they replace. Seedance 2.5 edits the source section, so the dance or action, camera move and timing follow the original closely. Fast spins, hands and very different body shapes can still drift, so check the whole result.
Can I swap a product in an ad and keep the hands and camera?
Yes. Choose Product and add clear images of your product, ideally on a plain background from more than one angle. The hand movement, camera path and timing are kept. Tight finger grips, fast rotation and small label text are the hardest parts and may change for a few frames.
Can I turn a normal video into anime without re-animating it?
Yes. Choose Style and describe the look, such as hand-drawn anime or a 3D render; you can add example images. The subjects, composition and performance are kept and only the rendering style changes.
How long can the video be?
Upload a clip from 5 seconds to 5 minutes, up to 50 MB, then choose a 5, 10, 15 or 30 second section starting at any second. Seedance 2.5 supports all four lengths; MiniMax H3 supports up to 15 seconds.
Does it support 1080p and vertical video?
Yes. Seedance 2.5 outputs 480p, 720p or 1080p and MiniMax H3 720p or 1080p. The result keeps your clip’s shape, snapped to 16:9, 4:3, 1:1, 3:4 or 9:16, so a vertical phone clip at 1080p comes back as 1080×1920.
How many credits does a Genjutsu AI video cost?
Seedance 2.5 at 720p costs 88 credits for 5 seconds, 175 for 10, 262 for 15 and 523 for 30. At 1080p a 10-second edit costs 310. MiniMax H3 at 720p costs 77 credits for 10 seconds. The generator shows the exact quote for every choice, and a failed generation returns its credits.
Is there a watermark?
Accounts without a paid order get a moving site watermark. The clean original is stored privately, and the same video downloads without the watermark once you have bought any plan or credit pack.
Which model should I choose?
Start with Seedance 2.5: it edits your clip directly and keeps motion and framing closest to the source, up to 30 seconds. Try MiniMax H3 for shorter clips when you want a bigger transformation and accept looser timing.
Keep the take. Change the world in it.
Pick a clip, add your references and see the edit in a few minutes. The quote is shown before you generate.
Open the Genjutsu AI editor