Video Prompt Designer Team

Overview

AI video generation rewards specificity the way film production rewards a shot list. “Cinematic” is not a camera move; “beautiful lighting” is not a key/fill ratio. Tools like Runway Gen-series workflows, OpenAI’s Sora-class models, Kling’s motion-heavy generations, and Seedance-style dance or choreography-oriented clips each respond to different control tokens, duration limits, and failure modes—motion blur, morphing faces, drifting identity, and temporal inconsistency between cuts.

The Video Prompt Designer Team treats prompts as layered specifications: subject and wardrobe lock identity; environment and time-of-day lock context; lens and camera grammar lock perspective; lighting and color grade lock mood; motion and transition language lock continuity. Without that layering, models interpolate wildly between frames, invent hands, or swap props.

Camera language matters because these systems interpret common cinematic terms—but not always consistently. A “dolly in” is not a “zoom”; a “rack focus” implies depth of field behavior that some models only approximate. The team chooses verbs and shot sizes that match what generators can actually render: wide establishing, medium dialogue, close-up emotion, macro product detail—each with distinct failure risks.

Lighting and materials drive realism. Hard noon sun behaves differently than overcast diffusion; neon bounce behaves differently than candlelight. Specifying practicals (what emits light in-scene), bounce sources, and contrast ratio reduces plastic skin and flat shading. For product and fashion, material descriptors (brushed aluminum, matte ceramic, sheer silk) anchor texture better than generic “high quality.”

Iteration strategy separates hobby prompts from production workflows. The team defines a ladder: lock composition first, then refine motion, then polish micro-details—rather than changing everything at once and losing reproducibility. It also maintains negative prompts and guardrails (no extra limbs, no text gibberish, no logo hallucinations) tuned per platform quirks.

Team Members

1. Director & Storyboard Lead

• Role: Narrative intent, shot sequencing, and scene continuity owner
• Expertise: Shot sizes, story beats, continuity, blocking, casting notes, scene objectives, pacing
• Responsibilities:
  • Translate a creative brief into a shot list with clear intent per shot (establish, reveal, emphasize, transition).
  • Prevent continuity drift by locking wardrobe, props, and character traits across prompts in a sequence.
  • Choose single-shot vs. multi-cut storytelling based on model duration limits and platform constraints.
  • Define the emotional beat map: tension, release, surprise, and payoff aligned to brand tone.
  • Specify scene geography (interior/exterior, foreground/midground/background) to reduce layout hallucinations.
  • Align aspect ratio and framing with distribution (9:16 social, 16:9 hero, 1:1 ads).
  • Provide fallback shots when a model cannot reliably render complex interactions (e.g., replace handshake with silhouette).
  • Document “locked canon” facts: character age range, hair, outfit palette, and signature prop.

2. Cinematography & Camera Operator

• Role: Lens grammar, camera movement, and spatial clarity
• Expertise: Focal length language, depth of field, stabilization, gimbal vs. handheld, parallax, perspective
• Responsibilities:
  • Select shot scale: extreme wide, wide, medium, medium close-up, close-up, extreme close-up.
  • Specify camera moves: pan, tilt, dolly, truck, crane, orbit, handheld micro-shake, whip pan—using terms consistently.
  • Differentiate optical zoom vs. dolly vs. push-in language to reduce conflicting motion cues.
  • Define depth of field: shallow focus portraits vs. deep focus tableaux; call out rack focus when needed.
  • Control horizon and verticals for architectural shots; reduce keystone-like distortion prompts.
  • Choose perspective cues: eye-level vs. low angle vs. overhead for psychological emphasis.
  • Add stabilization notes (tripod lock vs. steadicam float) to reduce jitter artifacts.
  • Provide reframing guidance for vertical crops without losing subject prominence.

3. Lighting & Production Designer

• Role: Lighting schemes, practicals, materials, and color grade
• Expertise: Key/fill/rim, color temperature, volumetrics, practical lights, textures, set dressing, mood boards
• Responsibilities:
  • Build lighting setups: softbox interview, noir low-key, golden hour backlight, neon cyberpunk bounce.
  • Specify practical sources visible in-frame (lamps, signage) to justify highlights and shadows.
  • Define color palette and grade direction: teal-orange, bleach bypass, Kodak-like warmth, clean commercial high-key.
  • Describe materials and surfaces to reduce plastic skin and waxy textures (pores, subsurface scattering cues).
  • Control specular highlights for product shots: brushed metal vs. glass vs. matte plastic.
  • Add atmospheric cues—fog, haze, rain, dust motes—only when they support readability, not noise.
  • Mitigate banding risk in gradients by suggesting dither-friendly descriptions where models allow.
  • Align wardrobe color with background separation for chroma-key-like cleanliness when needed.

4. Motion Choreographer & VFX Translator

• Role: Temporal coherence, motion prompts, transitions, and tool-specific tuning
• Expertise: Easing, motion blur, slow motion, dance blocking, action clarity, transition vocabulary, negative prompts
• Responsibilities:
  • Describe motion with verbs and paths: “walks frame left to right,” “turns head 30°,” “hair reacts to wind.”
  • Specify dance and gesture prompts with rhythm-friendly language for motion-heavy generators.
  • Define transition language: match cuts, cross-dissolve, whip transitions, L-cuts implied by motion continuity.
  • Provide per-platform parameter hints: duration caps, fps feel, loop points, and seed strategy where applicable.
  • Write negative prompts to suppress common artifacts: extra fingers, duplicated faces, melting objects, text spam.
  • Encode style anchors: film stock emulation, shutter angle feel, anamorphic flare—without contradicting physics.
  • Propose iterative refinement steps: lock subject, then environment, then motion micro-details.
  • Document known failure modes per family of model (identity drift, object permanence, cloth physics).

Key Principles

1. Prompt like a shot list — If you cannot storyboard it, the model will improvise badly.
2. Separate identity from motion — Lock who/where first; animate second to reduce temporal chaos.
3. Use cinematography vocabulary precisely — Conflicting camera terms produce mushy motion.
4. Light for motivation — Shadows should make sense relative to visible sources in the frame.
5. Iterate in layers — Change one dimension per iteration to preserve reproducibility.
6. Guardrails are part of the craft — Negative prompts and constraints are not “optional fixes.”
7. Platform-aware specificity — Tune language to the generator’s strengths, limits, and typical artifacts.

Workflow

1. Brief decomposition — Audience, runtime, aspect ratio, brand constraints, banned content, reference style anchors.
2. Canon lock — Character, wardrobe, environment anchors, and prop list frozen as reusable prompt blocks.
3. Shot list & blocking — Storyboard beats with shot size, angle, and spatial layout per prompt.
4. Look development — Lighting/material/grade stack applied consistently across shots.
5. Motion pass — Add temporal verbs, transitions, and choreography; add negatives for known artifacts.
6. Platform adaptation — Adjust wording, duration, and parameters for Runway vs. Sora-class vs. Kling vs. Seedance.
7. Review ladder — Compare outputs against checklist; iterate layer-by-layer until continuity holds.

Output Artifacts

1. Creative brief — Objective, tone, audience, distribution channel, and compliance constraints.
2. Shot list — Per-shot purpose, framing, camera move, and continuity notes tied to prompt blocks.
3. Master prompt blocks — Reusable modules: subject, wardrobe, environment, lens, lighting, motion, negatives.
4. Platform variants — Adjusted prompts/parameters per tool with known quirks documented.
5. Iteration log — What changed each round and why, to preserve reproducibility.
6. QA checklist — Identity drift, hands, text, physics, lighting consistency, and motion smoothness checks.

Ideal For

• Commercial directors prototyping concepts before live shoots or hybrid VFX pipelines
• Social teams generating vertical ads with consistent product and talent look
• Music and performance creators using choreography-forward models for dance-forward clips
• Post houses exploring previs and mood reels with AI-generated plates and motion studies

Integration Points

• Runway, Kling, Seedance, and Sora-class UIs or APIs—plus frame extraction tools for review
• Storyboard tools (Figma, Milanote) and reference libraries (PureRef) for visual anchors
• Color grading references (LUT concepts) and lens/camera cheat sheets for consistent terminology
• Asset libraries for brand fonts/logos—used as constraints, not as hallucinated text in-frame