LOADING
WHAT IT DOES
A face in scene one is the same face in scene fourteen.
Most tools re-roll the character every shot, because every shot is a fresh throw of the dice from the same words. Here a character is cast once, approved as a picture, and every later shot is filmed from that picture.
- 8
- poses in an identity sheet
- 1
- approved picture behind every shot
- ∞
- films that can reuse a cast
The look is a document, not a prompt
Each character carries a written descriptor — build, hair, wardrobe — that you can read and edit. It can be locked, so nothing downstream rewrites it, and it is versioned, so a descriptor that changed says which version it is on.
- Editable it is a field of text, not a hidden instruction.
- Lockable locked, no later step may touch it.
- Versioned and the version is on the screen.
You approve the picture before anything is filmed
The descriptor produces candidate renders and you pick the canonical one. From it comes an eight-pose identity sheet — front, three-quarter, profile, back, and the expressions — and those are what later shots reference.
- Eight poses so a shot from behind has something to work from.
- Filmed from the picture not from the words, which is where drift comes from.
- Kept a cast you approved is there in your next film.
Or use your own face
Upload a photo and it is cast exactly the same way: it becomes the canonical render, the identity sheet is drawn from it, and every scene is filmed from those. You were in a studio for none of it.