Style

The visual identity of this project — feeds every AI image prompt in the app.

NOT SAVED YET
This project has no Style manifest yet — nothing has been saved to stl/stl.json. Until it is, every AI image prompt in the app uses the built-in default: “painterly illustrated animation style, no photorealism”. Write the look you want below — it saves as you type.
SOURCE: stl/stl.json

The reusable phrase injected into any AI image prompt across the app. Keep it neutral — describe the look you want, not what to avoid.

This project has no Style manifest yet — nothing has been saved to stl/stl.json. Until it is, every AI image prompt in the app uses the built-in default: “painterly illustrated animation style, no photorealism”. Write the look you want below — it saves as you type.

Sent with every styled-video pass. @Video1 is the previs (motion, timing, who is on screen when), @Image1 the style frame (look only), @Audio1 the dialogue track. One per video mode — the mode is picked per shot on the beat board, and each one ships its own wording. Leave a box empty to keep the built-in default shown in it — those are load-bearing, so change them deliberately.

Follow

Previs exact — its motion, timing and every character's position, frame for frame. Right when the animation is finished or mocap.

Guide

Previs governs staging, camera and where everyone is; the model finds natural gait and weight. Right when the previs is positional blocking.

Prompt no previs sent

No previs at all — the style frame and the written direction alone. The clip is not sent.

Tween two pinned frames · no @-tokens

Inbetweens this beat's style frame and the NEXT beat's, in board order. Right when the beat is a key pose, not a shot — it carries timing, not staging. No clip, no audio, and no second picker.

The lip-sync instruction is composed on top of these automatically on any shot that has dialogue audio — you never need to write it here.

The kind of film — the structured grounding. These pick the look and tone, and are folded into the style spine sent to every AI image prompt.

Scoped to the production format.

The canvas frames are generated in — outputs are cropped from it.

The default era characters' wardrobe inherits — overridable per character.

Ordered — the first row is the primary. Each is cropped from the working aspect. Resolution is the shorter axis (Seedance ref-to-video caps at 720p).

Saves automatically as you edit — written to msd/stl/stl.json.