The duet, steps and movement

Make a Trouble AI video from two portraits

Your portrait

Start with two clear adult portraits

Choose a clear portrait of each consenting adult with the face and hair visible. Your photos guide the two identities in the dance. Appearance, clothing and details can vary in the finished video. Trouble AI uses two separate portraits: the first person and the second person are given distinct positions in the reference-guided duet. Use clear, well-lit photos with faces and hair visible, and avoid sunglasses or another person covering the face. A full-body photo can provide more information about clothing and proportions, while a close portrait gives a clearer view of the face. The finished scene is generated, so either kind of input may lead to changes outside the visible area. The example uses fictional adults; the two source images shown on this page are the inputs used for that example. Use only your own image or a photo of an adult who has agreed to the edit. Check the uploaded photos before starting, because each new video uses the displayed credits.

Try it now

The action

Follow the duet dance

The reference guides the pair from a face-to-face lean into synchronized steps and a front-and-back dance. Generated gestures, timing, music and framing can vary. This is a fictional AI remake. Trouble AI follows the broad shape of the Trouble Maker partner routine: the two performers start close together, move through coordinated steps, and change their relative positions as the dance continues. The reference gives the model a target for movement and camera framing rather than creating a frame-by-frame copy. Hands, feet, expressions and timing may differ, especially when the two people overlap. The example above is a real generated sample from two fictional adult portraits; play the whole clip to see how the identities and dance look throughout. It is not a recording of those people performing the dance. Sound is generated and can differ from the reference, so the original song or a particular performance is not guaranteed.

Try it now

The result

Review and save the video

The preset requests a fifteen-second vertical video. Review both faces, hands and movements when it is ready. Download your completed result or return to it in your creations. Downloading the existing file does not start another generation. Trouble AI creates a new fictional performance for each request, and there can be changes in appearance between frames. Watch the entire result and make sure both people remain recognizable, particularly during close contact and when one dancer moves behind the other. Check the hands, clothing, feet and camera framing as well. The still picture here is the second fictional adult photo used for the example; it is shown so you can compare the input with the finished clip. A new attempt is a separate generation with its own displayed credit cost. The existing completed video can be downloaded again without submitting another request. Processing includes provider work and result delivery, so completion time varies with demand; an active request can remain visible in your creations while it finishes.

Try it now