raindanceaiSign in

The Raindance AI prompt.
Already taken care of.

The studio applies this prompt on the server to every new duet. It combines your two identity photos with your selected 10-second reference clip, and preserves its composition in a padded 16:9 frame.

Treat reference Video 1 as the exact edit source, not inspiration for a new performance. Make only these identity replacements: Image 1 supplies the male identity and replaces the original male performer; Image 2 supplies the female identity and replaces the original female performer. Match each uploaded person's face, hair and recognizable identity while retaining the original performer's clothing, body pose and role in every shot. Track the original male and female by identity across cuts, movement, occlusion and changes of screen position. Never assign roles by left or right. In a female-only close-up, replace only the female; in a male-only shot, replace only the male. Do not add the other person to a solo shot. Keep the exact source shot order, cuts, shot durations, framing, camera position and movement, gestures, dance movements, lip movements, expressions, interactions and speed. Preserve the original setting, background, props, lighting, colors, composition, original outfits and all non-person elements. Do not restage the scene, create new choreography, add transitions, beautify or redesign the background, or invent any new shot. Do not retain the original performers' faces, merge the uploaded identities, swap genders or add extra people, text, watermarks or logos. Use only the source audio if supported; do not invent music, lyrics, dialogue or voices.

Why the reference video matters

A prompt alone cannot specify every gesture, camera movement and cut. The actual video supplies the performance. The photos supply the two people. We do not substitute an unrelated video or pretend a still image is a motion reference.

Keep the cast consistent

Image 1 always supplies the male identity; Image 2 always supplies the female identity. The prompt follows the original performer across cuts, movement and position changes. Female-only and male-only close-ups stay solo shots. Original outfits, poses, scene details, camera moves and shot timing are retained; only the person’s recognizable identity is replaced.

One prompt, three reference parts

The same prompt works with Part 1 (0:00–0:10), Part 2 (0:10–0:20) or Part 3 (0:20–0:30). The actual reference clip changes with your selection. Writing a timestamp into a prompt is not a substitute for supplying the correct video section. Keep the same male and female role assignments in every part.

Use the step-by-step guide to generate all three parts and join them in your editor.

What this prompt cannot promise

AI can still alter faces, clothing, timing and background details. A built-in prompt reduces setup work; it does not guarantee exact reproduction. The source preview includes audio; output audio fidelity is not guaranteed and music rights are not granted.

Use the built-in prompt