Scene Shot Prompt Grammar

# Scene and Shot Prompt Grammar

Use this file when compiling stylized scene plates, derived scene views, storyboard frames, cinematic keyframes, or when a user provides a prompt-format reference and asks to unify scene prompt style.

This grammar is a prompt-body layer only. It does not replace asset extraction, stable IDs, scene continuity cards, canvas metadata, QA notes, or model parameters.

## Core Order

Compile `SCN-*`, `SCN_VIEW-*`, and `SHOT-*` prompt bodies in this order:

1. `SHOT TYPE / VIEW`: shot scale, camera angle, view role, and aspect/framing when relevant.
2. `SUBJECT & ACTION`:
   - for `SCN-*`: the subject is the empty location plate; action becomes spatial state only, not character action.
   - for `SHOT-*`: one clear visual center plus character action or story beat.
3. `ENVIRONMENT`: geography, time, weather, set zones, thresholds, foreground/midground/background, fixed anchors.
4. `COMPOSITION RULES - fixed`: composition choices that must not drift: symmetry, one-point perspective, negative space, foreground obstruction, depth staging, frame-within-frame, leading lines, camera-parent relation.
5. `IMMUTABLE STYLE KEYWORDS - highest weight`: the locked LOOK layer, style_strength, medium anchors, palette behavior, light behavior, material system, grain/texture, halation or haze if used.
6. `NEGATIVE / LIMITS`: concise exclusions targeted to the asset type and model.

Do not put execution notes, canvas node data, rerun advice, or QA text inside this prompt body.

## Unified Scene Plate Format

Use this skeleton for `SCN-*` and `SCN_VIEW-*`: