Rumpelstiltskin AI Video Generator

Cast the dancer. Choose who watches.

A tiptoe performance in a straw-filled barn, with your choice of two characters. Upload a dancer photo and a spectator photo; the built-in video guides the gestures, reactions and changing camera shots.

Create my barn dance

Straw barn · built-in performance reference

This is the built-in motion reference, not a result generated with your photos. It is included in the generation request. Its ten-second footage has no audio track; an existing song is not included.

Make your Rumpelstiltskin AI video

Choose two different identity photos. Photo 1 replaces the dancer and photo 2 replaces the spectator across the changing shots. Neither photo is a first or last frame.

Upload the two characters

One clear adult per photo. JPG, PNG or WebP, up to 10 MB each. Use photos you have permission to use.

Photo 1 · Dancer

Choose a photo

Photo 2 · Spectator

Choose a photo

Editable performance prompt

Edit @Video1 into a ten-second fairy-tale barn performance. @Image1 is the DANCER: replace the short dancing man, including his face, hair, skin tone, body appearance, clothing and accessories. @Image2 is the SPECTATOR: replace the watching woman with the identity, hairstyle, body appearance, clothing and accessories in this second image. These are separate identity references, NOT first or last frames. Keep the dancer and spectator assigned consistently through every cut and viewpoint; their roles do not switch when their screen positions change. Use @Video1 for the straw barn, framing, lighting, camera movements, shot changes and timing. Follow its approach, finger-to-lips gesture, spectator reaction, tiptoe footwork, arm gestures and shoe close-up in the observed order. Do not introduce new choreography, extra people, mixed faces, wardrobe flicker, extra limbs, captions or logos. Do not add an existing song or recording. Prioritize coherent anatomy and stable identities over exact pixel matching.

MiniMax H3 · 768P · 10 seconds · reference framing · no song supplied

Cost: 460 site credits

Your video task

Your video and task number will appear here after submission.

My Creations

Refresh checks the existing task; it does not generate again. My Creations keeps queued, in-progress and failed records as well as finished videos.

How to make the two-character dance

Assign the dancer and spectator
Use one photo for each role. A visible face and an unobstructed outfit give the model clearer identity and clothing references. Swap the slots if the roles are reversed.
Keep the reference action intact
The editable prompt maps the dancer to image 1 and the spectator to image 2. Keep these roles and the reference-video instruction; add a small expression or scene note rather than a different dance.
Check the quote and follow the task
Submit once after checking the displayed credits. Save the task number and follow its progress here or in My Creations. Refresh checks the same task and does not create another paid video.

A dance and a reaction, in the same scene

The appeal is the contrast between a playful tiptoe dancer and the person watching. The preset uses a ten-second barn performance to guide the approach, shushing gesture, reactions, footwork and close-ups. It changes the cast rather than constructing a new scene from a text description alone.

Identity photos are separate from the motion reference

Your two photos guide faces, hairstyles, body appearance, clothing and accessories. The built-in video guides the barn, shot changes and action timing. Characters may change screen positions; they keep their dancer and spectator roles. Identity transfer and fine details can vary.

Review both characters across every cut

Watch for mixed faces, clothing changes, extra fingers and role switches, especially in close-ups. Compare the dancer's footwork and the spectator's reaction with the reference. This page has not been verified by a new paid generation; matching every frame is not guaranteed.

Rumpelstiltskin AI video questions

How many photos do I need?

Two different photos: one for the dancer and one for the spectator. These are identity references, not the opening and ending frames. The motion video is already built in.

What happens to outfits and the scene?

The prompt asks for each uploaded character's face, hair, body appearance, outfit and accessories. The reference supplies the straw barn, lighting, gestures and cuts. This is a requested edit; precise likeness or unchanged details are not guaranteed.

Does it include the trending music?

No existing song or recording is supplied. The built-in ten-second reference is silent. Add music separately using audio you are entitled to use; soundtrack preservation is not promised.

How are credits and failures handled?

The quote includes a ten-second output, ten seconds of reference video and two identity images under the site's current pricing rules. Credits are reserved on submission. Confirmed failures follow the existing refund rules; a status refresh or browser reload does not create another paid task.