skill: veo-cinematic-reels

Name: Veo Cinematic Reels — scene scripts + Google Flow video prompts Description: Turns a reel idea into two deliverables that work together: a scene-by-scene script (hook → beats → payoff, with timings) and a copy-paste prompt pack for Google Flow / Veo 3.1 — one cinematic video-generation prompt per scene that the user pastes manually, generates, and stitches into a reel. The prompts are engineered so the character never changes between scenes: a reference-image ingredient (uploaded once) plus a verbatim character block embedded in every single prompt, along with IMAX-scale camera language, a locked color grade, and native dialogue/SFX audio. Built for creators who generate scenes one at a time in Flow's Ingredients→video or text-to-video panel.

The quality bar (non-negotiable — read before anything else)

Rail	Rule
Character consistency (the #1 rail)	The same character across every scene. Enforced TWO ways, both mandatory: (1) the user uploads the character reference-image ingredients (from the character sheet, § Workflow Stage 2) into Flow's Ingredients panel, AND (2) the exact same character block text appears VERBATIM in every scene prompt — never reworded, never trimmed. The script self-verifies this; if a prompt misses a word, regenerate it.
One prompt per scene, copy-paste ready	Every scene gets ONE complete Veo prompt following the 5-part formula (cinematography + subject + action + context + style), with the consistency tokens appended. No instructions, no meta-commentary inside the prompt — the user copies and pastes it as-is into Google Flow.
IMAX-level cinematic language	Camera language is explicit and large-format: camera motion (dolly/tracking/crane/push-in/arc), lens + focal length, depth of field, framing (wide/medium/close-up/ECU), and the IMAX token block (below) — never vague "cinematic shot". Vertical 9:16 for reels (Veo 3.1 native aspect).
Color grading locked per reel	ONE grade token block (film stock + palette + contrast/grain) chosen at the start and appended verbatim to every prompt — teal-and-orange, warm Kodak, cold sci-fi, etc. Never two grades in one reel.
Native audio in the prompt	Veo 3.1 generates dialogue + SFX natively. Dialogue is written inside quotes, SFX/ambience labeled `SFX:`/`Ambient:` in the prompt, and delivery (tone/pace) specified — so the voice and sound design come out of the generation, not post.
Scenes stitch into a reel	5–9 scenes, each 4–8s (Veo clip lengths), total 30–60s. Scenes follow the retention arc: hook (0–3s) → agitate → payoff → CTA/loop. Lighting and location continuity are written across scene boundaries (Scene N ends golden-hour → Scene N+1 starts with the same warm low light).