ai-video-storyboard

AI Video Storyboard Generator

Overview

AI video generators produce 5–15 second clips. Real-world videos are longer: 30s TikToks, 60s ads, 90s explainers. This skill bridges that gap by producing a complete shot-list storyboard — a coordinated sequence of per-shot prompts with shared visual language, so the assembled final video looks like one intentional piece of work rather than six disconnected clips.

Core insight: Visual consistency across shots matters more than any single shot being perfect. A mediocre but consistent set of shots edits together; six gorgeous but mismatched shots do not.

When to Use

• User wants a video longer than 15 seconds
• User wants a TikTok / Reel / Short / ad / explainer / B-roll sequence
• User says "help me make a video about X" and has no shot list
• User is about to generate multiple clips and needs a plan

Do NOT use for:

• Single-shot generation (just help with one prompt instead)
• Static image generation
• Video editing advice after clips are generated (that's a separate concern)

Workflow

Step 1 — Brief Intake

Ask the user these questions in a single message. Accept short answers:

1. Goal platform and duration? (TikTok 30s / Reel 15-60s / YouTube Short / Instagram Ad / explainer)
2. What is the video about? (subject, story, product)
3. Brand vibe / tone? (cozy, energetic, premium, minimalist, playful, cinematic, etc.)
4. Call to action at the end? (visit website, buy product, follow channel, etc.)
5. Any hard constraints? (must include logo, specific colors, locations, etc.)

If user already provided some of these, skip those questions and confirm the rest.

Step 2 — Infer Structure

Based on duration and platform, divide the timeline into shots of ~5 seconds each (the sweet spot for AI video generators).

Standard cadences:

Step 3 — Establish Visual Consistency Layer

Before writing any shot, lock in the shared visual language. This is what makes shots edit together:

• Color palette (3-5 specific hex values or named colors)
• Lighting style (golden hour / neon / overcast / motivated / cinematic rim light)
• Lens character (shallow DOF / deep focus / wide angle distortion / macro)
• Film look (clean digital / 16mm grain / anamorphic / VHS / 35mm)
• Motion language (handheld / locked off / dolly only / gimbal smooth)

Write these as a Visual Theme block at the top of the output. Every shot's prompt must respect this block.

Step 4 — Write Each Shot

For each shot, produce this structure:

## Shot N (START-ENDs) — [Purpose label: Hook / Setting / Action / Detail / Reveal / CTA]

**Composition:** [shot type + angle, e.g., "Extreme close-up, overhead"]
**Camera move:** [locked / slow dolly in / tracking / crane up / etc.]
**Lighting:** [from the Visual Theme, applied to this scene]
**Subject:** [what is in frame]
**Action:** [what is happening]

**Prompt to copy:**
> [Complete, cinematic-quality prompt, 40-80 words, including: subject + action + environment + camera + lighting + style + technical spec (duration, aspect ratio, resolution). Always ends with "cinematic 1080p, synchronized audio"]

Critical rules for shot prompts:

1. Always specify the shared visual language in every prompt (color palette, lighting, lens character, film look). This is how you enforce consistency.
2. Specify exact duration (4s / 5s / 8s) and aspect ratio (9:16 for TikTok/Reels, 16:9 for YouTube landscape, 1:1 for feed posts).
3. Always end with "cinematic 1080p, synchronized audio" — this signals professional-grade quality and works with modern video models that support both.
4. Use cinematography vocabulary: ECU (extreme close-up), CU, MS (medium shot), WS (wide shot), OTS (over-the-shoulder), POV, Dutch angle, low angle, high angle, bird's eye, dolly in, dolly out, tracking, handheld, rack focus, etc. See references/shot-types.md for the full vocabulary.
5. Don't write abstract prompts — be concrete. "A woman" → "A barista in her late 20s with wavy auburn hair, wearing a denim apron".
6. No model-specific hacks — don't write prompts tuned to specific model quirks. Write model-agnostic cinematic prompts that work anywhere.

Step 5 — Add Narrative Structure

The sequence of shots must have a story arc, not a random list. Use one of these patterns:

Pattern A — Hook / Build / Payoff / CTA (TikTok default)

• Shot 1: Visual hook (stop the scroll)
• Shot 2-3: Build context / intrigue
• Shot 4-5: Main content / payoff
• Shot 6: CTA

Pattern B — Problem / Solution / Proof / CTA (Ad default)

• Shots 1-2: Relatable problem
• Shot 3: Your product as solution
• Shots 4-5: Benefits / results
• Shot 6: CTA

Pattern C — Atmosphere → Climax (Brand story)

• Longer atmospheric shots
• Slow reveal
• Emotional climax
• Logo reveal

Step 6 — Add Post-Production Checklist

Close with actionable post-production notes:

Step 7 — Explain Why It Works

Close with a "Why this works" block explaining the creative decisions. This educates the user and differentiates your output from generic prompt lists. Reference:

• The hook rule (first second determines watch-through)
• Pacing cadence (average scroll time)
• Story structure (why the shot order matters)
• Platform-specific conventions

Output Format

The final output is a single Markdown document containing:

1. Header — Title, duration, shot count, aspect ratio
2. Visual Theme — Shared palette, lighting, lens, film look, motion
3. Shot List — N shots, each following the structure in Step 4
4. Post-Production Checklist
5. Why This Works — creative rationale

See examples/ for four complete sample outputs:

• tiktok-reel-30s-coffee.md — Coffee shop opening TikTok
• (more examples to be added)

References

• references/shot-types.md — Cinematography vocabulary (ECU, CU, MS, WS, OTS, POV, angles)
• references/camera-moves.md — Camera movement vocabulary (dolly, crane, pan, tilt, tracking)
• references/lighting.md — Lighting terminology (golden hour, rim, motivated, practical)
• references/genre-templates.md — Pre-built templates for common genres (food, fashion, tech, travel, SaaS)

License

MIT — use freely, commercial or personal.