Genjutsu AIAI video creation, built around your references.

Bring a character photo and a performance video. Recreate the motion, explore a new look, and make the scene your own.

Start creating Explore tools

Motion TransferObject SwapPhoto + video references

Concept artwork · not a generated video

MiniMax H3 · sign in to check generation access

Your character. Their moves.

Choose a workflow, prepare your references, then review the supported settings and cost.

Character photo

JPG, PNG or WebP · up to 10 MB

Performance video

MP4 or MOV · 2–15 seconds · max 50 MB

Creative starters

Describe your video110 / 1200

Advanced settings

Preferred duration

H3 uses the whole reference clip. Upload a 2–15 second clip; output duration is 4–15 seconds.

Output preferences

Preferred resolution

MiniMax H3 · 768P or 2K · one video per submission. Editing follows references and may also change surrounding details.

Do not upload someone else’s photo or video without their consent.

TWO WAYS TO CREATE

Keep the performance. Reimagine the shot.

Start with footage you already have. Choose whether to rebuild the scene or change a single element.

Motion Transfer

A new character. The same moves.

Prepare a source performance and a character photo. Describe the performer and setting you want to recreate around the original movement.

Use this template

Object Swap

Change one detail in the take.

Select a person, product or background as your target. Pair the original clip with a replacement reference and a focused prompt.

Use this template

CREATIVE STARTING POINTS

What will you make next?

Explore a direction and load its prompt into your workspace. Add your own references to make it yours.

Motion Transfer

Character recast

Give your original dance or performance a new identity. Start with a clear full-body character reference.

Workflow illustration · not a generated video

Use this template

From reference to performance.

1. Prepare the references

Use a clear character photo and a video with visible movements. Match the body framing where possible.

2. Review before generating

Check model support, duration, output settings and the credit quote before committing.

3. Inspect the full clip

Compare the face, outfit, movement and scene throughout the video. Download your finished result from your library.

Find your next performance.

Purpose-built setups for character videos, dance references and two-person performances.

AI Character Video Generator

Use your own performance clip and a character photo to prepare a reference-led recreation.

Open workspace

AI Dance Video Generator

Make your character follow a dance reference. Choose a clear full-body photo and upload a performance clip.

Open workspace

Hotel Lobby AI Video Generator

Make a Hotel Lobby AI video online from two photos or one duo photo. Built-in 15-second template, no prompt or reference upload needed. Sign in to generate.

Open workspace

Before you start.

What is Genjutsu AI video?

Genjutsu is a named Higgsfield workflow. This independent app prepares reference-led video tasks; each connected model is identified separately. It is not affiliated with Higgsfield.

Is motion transfer the same as object swap?

Motion transfer recreates a performance with a character reference and can rebuild the surroundings. Object swap aims to change a selected element in an existing clip. They require different models.

Can I use a photo alone?

Motion transfer needs a performance video. A tested template can supply one, but templates without a connected reference clip cannot generate.

Is generation free?

Preparing inputs is free. Supported generation shows a credit quote before submission. Free credits and purchase options depend on the account configuration.

REFERENCE-LED VIDEO CREATION

Your next video starts with one reference.

Choose a tool, explore the settings and prepare your references for your next video.

Start creating