DIY route
The Rumpelstiltskin AI prompt
Want to recreate the tiptoe barn scene yourself in a video-swap model? Here is the prompt, what it needs, and when the $9.90 shortcut wins.
Copy-ready prompt
Replace the person in the reference video with the person in my photo. Keep the scene, camera moves, choreography and 1980s dark-fantasy film look exactly as they are - a candlelit barn full of straw, hard cuts, grainy footage. The character tiptoes in, shushes the camera, grins widely, then breaks into a bouncing arms-out dance. Preserve my photo's face and hair; dress the character in a black velvet tailcoat.
You still need the source barn clip as a reference video and a model that accepts reference-video plus photo input (Seedance, Kling or Vidu class). Results vary a lot between runs - expect a few paid retries.
How to use it
- 1
Find the source clip
The prompt alone does nothing - these models swap faces inside a reference video. Save the original barn-scene clip (the one every generator uses) to use as your motion template.
- 2
Pick a video-swap model
Use a model that takes a reference video plus one or two photos - Seedance 2.0, Kling or Vidu class tools all offer this. Upload the clip first, then your photo, then paste the prompt.
- 3
Budget for retries
Face consistency, the tiptoe gait and the gold-ending beats are the parts that fail. Most DIY attempts need 3-5 renders before one looks right - each one costs credits on that platform.
DIY prompting
- Hunt down the reference clip yourself
- Tune the prompt until faces stop melting
- Pay for every failed render - no refund
- Handle the song and AI label yourself anyway
Rumpelstiltskin AI
- Template, prompt and cast notes already locked in
- From $9.90 a video - most people spend less than DIY retries
- Failed renders refund credits automatically
- Clean silent MP4, no watermark, kept in your library
Skip the prompt - one or two photos and about five minutes.
Make my video instead