Baby Got Back — Three-Photo Dance Video on a White Background
This template builds a short dance clip out of three photos instead of one. The second photo sets the person — face, hair and body stay as they are. The first and third photos set the outfits, and the clip moves between them. The background is replaced with a seamless white studio, which is what makes the result look shot for a shop rather than filmed in a bedroom.
Why three photos instead of one
A single photo can carry a person and an outfit, but not a second look. With three, the model has enough reference to keep the same face across a wardrobe change. Photo two is the anchor: it decides who is on screen. Photos one and three decide what they are wearing. Put the clearest face in the middle slot and clean, flat-lay or full-body outfit shots on either side.
What the motion reference does, and does not, do
The clip borrows movement and timing only. It does not borrow the look or the identity of the reference footage: a black-and-white dance clip stays a movement source, not a visual style, and the person in it never replaces the person in your photos. That separation is why the output keeps your model's face while still getting a real choreographed rhythm instead of a generic sway.
Getting a clean result
Shoot the person photo against a plain wall in even light, standing, full body or three-quarter. Outfit photos can be worn shots or flat lays, as long as the garment shape is clear and not cropped. The white background is generated, so nothing in your original room survives — but a cluttered backdrop still makes the cutout worse at the edges. One person per photo, no group shots.
Photo checklist
- Middle slot: sharp, front-facing photo of the person in even light.
- Side slots: the two outfits, full garment visible, no heavy cropping.
- Avoid sunglasses, hats that hide the hairline, and backlit windows.
- One person per photo — group shots confuse the identity anchor.
Not sure your photo will work? Read what photo works best for an AI video.
Questions
How many photos do I need?
Three. The second one is the person; the first and third are the outfits. All three are required before the generate button will run.
Can I use a photo of someone else as the motion reference?
Yes. The reference video only supplies movement and timing. Its person, its scene and its colour treatment are all discarded.
What if my outfit photos are cropped?
Cropped garments still transfer, but sleeves and hems get invented by the model. A full garment in frame is more faithful.
What does one video cost?
Ten credits, the same as any other video template. Image generations are five credits each.