A working talk dirty ai prompt has to solve six problems at once: keep only the lead from your photo, keep the crew, the stage and the flames the same, fit the whole routine into 15 seconds, stop the crew from copying your face, leave the sound for afterwards and start on a wide, dark opening shot. AI Talk Dirty is a dance video maker, not a chat app, and the prompt behind it is already written and tested. Make the video from one photo.
What the prompt has to solve
- Only the lead comes from the photo. The person, hair and outfit must match the photo. Everything else in the frame belongs to the stage and must not take anything from it.
- The crew, the stage and the flames stay fixed. The dancers in black-and-yellow jackets, the spotlight circles, the giant screen and the flame jets are what make the clip recognisable. A loose wording lets them change from one render to the next.
- Timing to 15 seconds. The routine has a fixed order: walk-on, lights, spotlight, flames, the crew, the finale. If a model is not held to it, beats get skipped or stretched and the finale arrives early.
- The crew must not copy the face. Without a clear instruction every dancer can drift toward the lead's face.
- The sound comes afterwards. The picture is rendered silent and the trend sound is mixed in at the end, so the prompt does not describe audio.
- A wide, dark opening. The first shot shows the lead small and in shadow. The wording has to survive that, or the face is invented from a few pixels.
Why copied prompts fail
A pasted prompt leaves out the stage reference and the inputs the original render used, so the model improvises: the crew gets the lead's face, the flames move and the routine runs short. The full case study is in five ways a copied dance-swap prompt breaks on stage footage.
What is built in here
AI Talk Dirty runs on Seedance 2.0 / 2.5 by ByteDance. You choose a tier, and the prompt behind it does not change:
- Standard: Seedance 2.0 Mini, from 5 credits per video.
- Fast: Seedance 2.0 Fast, from 8 credits per video.
- Pro: Seedance 2.0, from 10 credits per video.
- Ultra: Seedance 2.5, from 17 credits per video.
Each of the 6 stage looks is a fixed package, written once and tested, so every render follows the same shot list. You only upload a photo and pick a look. The prompt is not shown because there is nothing for you to edit. A failed or blocked render returns its credits automatically; see pricing for costs.
For the steps, read how to make the video; for the background, the trend explained.