Window Dance AI

Window Dance AI photo tips: the photo that actually works

November 3, 2026 · 4 min read

The prepared scene. Your photo is the one dancer in the window — and it is the only upload there is.

Everything about a Window Dance AI result comes back to one photograph. The clip has a single dancer in a single window, so that photo carries the face, the body, the clothes and the read on who is dancing — and the model has to hold all of it steady for twelve seconds of movement.

The four things that decide the result

1How much of the body is in the frame. Waist up or full body, not a crop at the chin.
2How sharp the face is. The face is the one thing people recognise, and it is the one thing the model has to keep.
3How even the light is. Hard shadows on one side of the face get carried into the clip.
4How many subjects are in the photo. Exactly one — a person, or a cat, or a dog.

Good photos, and what they have in common

What you see in the photo	Why it works
Waist-up shot, arms visible	The model can copy the outfit and the posture instead of inventing them
Face square to the camera	The face survives being turned and moved; a three-quarter view gets reinvented
Sharp, no heavy filter	Filters and smoothing are detail the model has to guess at, and it guesses wrong
Daylight or a plain lamp	Even light keeps the features readable through the whole clip
Plain background	A busy room gets rebuilt behind your subject and the edges wobble
One adult, or one animal	The clip has one dancer; extra faces appear beside yours

The mistakes that quietly ruin a render

What you see	Why	Fix
The face does not look like you	Small, blurry or filtered photo	Use a sharp, front-facing photo in even light and retry the preview
Clothes or hair change	The photo is cropped at the face	Use a waist-up or full-body photo so the outfit is visible
A stranger appears in the room	More than one person in the photo	Crop the photo to one adult (or one pet) before uploading
Hands pass through the window frame	Fast moves are hard for AI	It is a known AI limit; another preview try often helps

What to write under the photo

There is one optional line under the upload, and it is a note about the subject — not a prompt. It can change what the dancer looks like, and nothing else.

“A grey suit and a red tie.”
“A yellow raincoat.”
“Short blonde hair.”
“A green collar on the dog.”

Leave it empty and the dancer wears what your photo shows. Fill it in and it replaces that outfit. Either way the street, the window, the camera move and the timing stay exactly as they are in the prepared clip.

People and pets use the same rules

A cat or a dog photo is judged by the same four things: how much of the body is in frame, how sharp the face is, how even the light is, and how many animals are in it. One pet, one photo, facing the camera, in focus. Pet Window Dance has the specific notes for animals.

How long the render takes

Most renders land in five to eight minutes. You get an appearance preview first, and you can try it up to three times before paying for the full clip. If a render fails on our side the credit comes back automatically.

Window Dance AI questions, answered

What photo works best for a Window Dance AI video?

One person or one pet, facing the camera, face sharp and reasonably large in frame, with a plain background and even light. A waist-up or full-body shot works better than a tight face crop, because the crop is also what the model uses to guess the outfit.

Can I change what we are wearing?

Yes. Under your photo, write what the dancer should wear or look like — a white tank, a suit, a new hairstyle, a different coat on the pet. That instruction applies to that one photo alone.

What happens if my photo has two people in it?

A second face can end up in the frame next to yours. Crop the photo to one adult, or one pet, before uploading.

Does the AI version keep my face the whole time?

Your photo is the character reference, so the model keeps your face and body consistent through the clip rather than pasting a still photo on top. It is still a generated video, so details shift slightly between renders — that is the model working, not a defect.

What file types can I upload?

JPG, PNG or WebP, up to 10 MB. Large photos are compressed on the way in so that the upload survives a slow connection.

Make it yours

The street, the lit window, the camera move and the timing are already filmed in the clip above. Send one photo and the same window comes back with you in it.

Window Dance AI is one photo and one payment. The clip renders in about five to eight minutes.

Make my Window Dance

More guides