💛 @aiwithdonna About Me ✨ All Guides
AIGC Prompt Guide & Workflow

Swap Face & Background in a Dance Video

Any face, any city — the exact same walk-into-dance moves every time

Hey, it's Donna! 💛 This one's inspired by a street-dance reel that's been going around — original concept and choreography by @mynycvision — a girl walking down the street and just breaking into full, free, no-one's-watching dance moves. I recreated the entire camera setup shot by shot (distances, angles, cuts and all), swapped myself in, and rewrote the setting into a Parisian street with the Eiffel Tower in view. Drop in your own face + outfit reference and rewrite the background in the prompt to wherever you want to be, and you're the one dancing through it.

🎥 Original concept & choreography by @mynycvision
Demo Video

Reference Video & Download

Feel So Free in Paris (1080p HD, 11s)

Donna dancing down a Parisian street toward the Eiffel Tower — download this and use it as your motion + camera reference for the swap.

Download Reference Video
🎵
Song backup: If this video ever gets muted or taken down over the song's copyright, download the 11s audio here and add it back in yourself.
Concept

What are we creating?

A video where your face and outfit replace the dancer's, and the background becomes wherever you describe in the prompt — a different street, a different city, a different landmark entirely — while the exact same walk-into-dance choreography and camera moves (wide opening shot, push-in to medium, spin cut, cut back to front) stay untouched.
Visual Input

📌 Reference Photo

image_1 Face + Outfit Reference Example
Face + Outfit Reference (required)
This is the photo used to make the demo video above. Swap in your own front-facing photo here — face, hairstyle, and outfit all come from whatever you upload.
image_2 Background Reference Example
Background Reference (optional)
Want to dance somewhere else? Upload a photo of that place here. Leave this out and the default Paris street with the Eiffel Tower (shown here) is used instead.
⚡
image_2 is optional. Skip it and you get the default Paris background baked into the prompt. Add it and the prompt swaps in whatever location your photo shows.
Execution

🪜 Simple 3-Step Flow

Step 1
Pick your reference photo(s)
image_1 (required): your face, hairstyle, and outfit — any clear, front-facing, well-lit photo. image_2 (optional): a photo of wherever you want to dance instead. Skip it and you get the default Paris street.
Step 2
Upload the reference video
Download the video above and drag and drop it into the chat. It's a smart agent that understands what you tell it, so it reads the video's motion and camera work automatically.
Step 3
Drop in the prompt and generate
Drag and drop your photo(s) into the same chat, then paste the prompt below as-is. image_1 and image_2 map automatically — no manual edits needed either way.
⚡
Background is automatic: No image_2? The prompt defaults to the Paris street. Added image_2? It swaps in whatever location that photo shows.
Ready to Paste

🎬 The Full Prompt

⚠️ Read this before you paste
  • This is a swap, not a text-to-video generation. The prompt alone does nothing — you must also upload the reference video from the Demo Video section above into the same chat, along with your photo(s).
  • Only image_1 is required. You can paste the prompt exactly as written below — nothing needs to be rewritten.
  • The background is not described in the prompt on purpose — it comes from the video itself and stays untouched unless you add image_2.
  • The prompt says "the woman" / "she" because that matches the reference video and your photo — if your image_1 is a different gender, replace those words with the correct pronoun before pasting.
swap-face-background-street-dance-prompt.txt 
REFERENCE MAP

Image 1 = FACE + OUTFIT REFERENCE (required)
Use Image 1 as the exact identity reference — face, hairstyle, skin tone, and outfit.

Image 2 = BACKGROUND REFERENCE (optional)
Only needed if you want a different location. Leave it out and the video's own background is left untouched.

FACE + OUTFIT SWAP TASK

1. CHARACTER REPLACEMENT: Replace the woman in the video with the character from @[Image 1](image_1). Match her face, hairstyle, skin tone, body proportions, and outfit exactly as shown in Image 1. Keep her exact body position, pose, scale, walking path, and screen placement identical to the original video in every frame.

2. BACKGROUND: If @[Image 2](image_2) is provided, replace the entire environment — street, buildings, landmark, sky, everything visible behind and around her — with the location shown in Image 2, matching perspective, depth, and vanishing point to the original camera moves. If Image 2 is NOT provided, keep the background 100% identical to the source video — same street, same buildings, same Eiffel Tower, unchanged in every frame.

3. MOTION AND TIMING: Preserve the original choreography frame by frame — the same walk-in, hip sways, arm pumps, spin, and hair flip, at the exact same timing and pace. Do not add, remove, or re-time any movement.

4. CAMERA AND FRAMING: Keep the original camera work unchanged — the same wide opening shot, smooth push-in to a medium shot, cut to the behind-the-shoulder spin, and cut back to the front. Do not re-frame, crop, zoom, or add new camera moves.

5. LIGHTING INTEGRATION: If Image 2 is used, match the lighting direction, intensity, and color temperature of the new background onto the character, with correct contact shadows, so she looks natively filmed there, not composited. If Image 2 is not used, keep the original lighting untouched.

6. PROPS: Keep the phone in her hand and the wired earphones as in the original video, moving naturally with her.

7. IDENTITY LOCK: No morphing, no identity drift, no blending between her face/outfit and the original dancer's, in any frame.

AVOID: extra people added, warped or flickering face, mismatched lighting between subject and background, style change, cartoon or filter look, watermark, text overlay, slow motion or speed ramp.

OUTPUT: same resolution, same aspect ratio, same frame rate, same length as the source video.
Practical Tips

💡 Before You Try It

  • •
    Pricing & quota: New accounts get a short free run at Starwish — enough to test the swap before committing to a paid plan.
  • •
    Processing time: It still takes a few minutes to render — go do something else while you wait.
  • •
    Background photo choice: Pick a photo with a clear street, path, or vanishing point heading away from camera — it matches this video's straight walking line best.
  • •
    Photo quality: Use a front-facing, sharp, evenly lit photo for image_1. Clean backgrounds make identity isolation cleaner.
  • •
    Prompt importance: The prompt locks identity, background, and motion. Omitting constraints causes face drift and timing shifts.
Daily Creative Drops ⚡

Keep Creating With Me 🤍

If you use one of these prompts, tag @aiwithdonna so I can see what you create.

More guides coming soon!