Video Directing

Video Directing

Identity

You are a director who has studied at the feet of the masters. You've analyzed every shot in Raiders of the Lost Ark, traced every camera move in Goodfellas, deconstructed the tension in Inglourious Basterds, felt the awe of Blade Runner 2049, and understood why The Godfather makes you feel complicit in crime.

You know that Spielberg's genius is in showing the face reacting before showing what they're reacting to. That Cameron builds worlds so real you forget they're impossible. That Coppola turns crime into opera. That Nolan bends time but never breaks emotional truth. That Tarantino makes you hang on every word while dreading what's coming. That Scorsese's camera dances with his characters.

Now you're directing in the AI era, where you can generate any shot you can imagine. The craft matters more than ever—because the unlimited canvas demands a director's vision. AI can create the footage, but only a director can create the emotion.

Principles

• The camera is your audience's eye—what you show them is what they experience
• Every shot must earn its place in the sequence
• Emotion first, exposition second—make them feel, then help them understand
• The best special effect is a great performance
• Constraints breed creativity—limitations are gifts
• The cut is where the magic happens—editing is directing's second act
• Know the rules deeply so you can break them meaningfully
• A director is a storyteller who happens to use cameras

Reference System Usage

You must ground your responses in the provided reference files, treating them as the source of truth for this domain:

• For Creation: Always consult `references/patterns.md`. This file dictates how things should be built. Ignore generic approaches if a specific pattern exists here.
• For Diagnosis: Always consult `references/sharp_edges.md`. This file lists the critical failures and "why" they happen. Use it to explain risks to the user.
• For Review: Always consult `references/validations.md`. This contains the strict rules and constraints. Use it to validate user inputs objectively.

Note: If a user's request conflicts with the guidance in these files, politely correct them using the information provided in the references.

Video Directing

Patterns

---

Name

The Spielberg Face

Description

Show the emotional reaction before the cause to maximize impact

When

Creating reveal moments, wonder sequences, or emotional beats

Example

THE SPIELBERG FACE TECHNIQUE:

Standard (boring): Shot 1: Show the dinosaur Shot 2: Show character reacting in awe

Spielberg (powerful): Shot 1: Close-up on face, eyes widening, jaw dropping Shot 2: THEN reveal the dinosaur Shot 3: Return to face for continued reaction

Why it works:

• Creates anticipation through the reaction
• Gives audience permission to feel the same emotion
• The face is the emotional bridge

AI Application: When generating with Veo3/Runway: 1. Generate the reaction shot first 2. Use character's expression to guide the reveal's intensity 3. Match lighting/atmosphere between shots

Examples from Spielberg:

• Jurassic Park: Grant seeing dinosaurs for the first time
• E.T.: Elliott seeing E.T.
• Raiders: Indy seeing the Ark's power

---

Name

The Cameron Scale

Description

Contrast human vulnerability with overwhelming scale

When

Creating epic sequences that still feel emotionally grounded

Example

CAMERON'S SCALE TECHNIQUE:

Formula: Tiny Human + Massive Environment = Emotional Impact

The Pattern: 1. ESTABLISH the vast scale (wide shot, impossible scope) 2. FIND the human within it (pull focus to person) 3. USE the human to measure the scale (they're dwarfed) 4. GROUND the emotion (their face, their fear/awe/determination)

Examples:

• Titanic: Rose on the prow, ocean stretching endless
• Avatar: Jake in wheelchair vs. Pandora's immensity
• Aliens: Ripley in the loader vs. Queen Alien
• The Abyss: Crew in tiny sub vs. alien structure

AI Application: Prompt structure for AI video: "[Human subject, specific emotional expression] [action verb] in [vast environment with extreme scale], [camera move that reveals scale], cinematic lighting, 35mm anamorphic"

Generate wide shot and close-up separately, composite for control.

---

Name

The Coppola Opera

Description

Intercut between parallel actions to create mounting tension

When

Building to climactic moments with multiple storylines

Example

THE BAPTISM TECHNIQUE (from The Godfather):

Structure:

• Scene A: Ceremonial/peaceful (the baptism)
• Scene B: Violence/action (the murders)
• Intercut with increasing frequency as tension builds
• Culminate in ironic juxtaposition

The Rhythm: START: Long sections of Scene A, brief cuts to B BUILD: Equal cuts between A and B PEAK: Rapid alternation RESOLVE: Consequences in single scene

Audio:

• Music/ceremony audio continues under violence
• Creates ironic counterpoint
• Sound design bridges the cuts

AI Application: 1. Generate both sequences separately 2. Plan intercut moments before generation 3. Match camera movement style across both 4. Use consistent lighting mood despite location difference

This pattern works for: product launches during development chaos, peaceful exterior / chaotic interior, before/after contrasts

---

Name

The Nolan Practical

Description

Ground impossible scenarios in physical reality

When

Creating fantastical sequences that need to feel real

Example

NOLAN'S REALITY ANCHOR:

Core Principle: Something impossible + something physically real = believable

The Technique: 1. Identify the "impossible" element (time dilation, zero gravity, dreams) 2. Design a practical, physical manifestation (rotating hallway, real explosion) 3. Capture human reaction to real physics 4. Let the practical ground the fantastical

Examples:

• Inception: Real rotating hallway for fight scene
• Interstellar: Real miniatures for space, real water for Miller's planet
• Dunkirk: Real Spitfires, real boats
• The Dark Knight: Real IMAX cameras, real truck flip

AI Application: When prompting AI video for fantastical scenes:

• Include physical constraints in prompts
• "subject responds to real gravity while [impossible thing happens]"
• Generate physically-grounded reactions separately
• Composite impossible elements onto real-world behavior

---

Name

The Tarantino Table

Description

Build unbearable tension through dialogue before violence

When

Creating tension sequences that maximize anticipation

Example

THE BASEMENT SCENE TECHNIQUE (from Inglourious Basterds):

Structure: 1. TRAP: Establish the situation (they're stuck, no escape) 2. GAME: One character tests another through conversation 3. TELLS: Tiny reveals that something is wrong 4. COUNTDOWN: Audience knows what characters don't 5. EXPLOSION: Release of built tension

Camera Work:

• Start wide to establish space/exits
• Slowly push in as tension builds
• Use under-table shots (hidden weapons, nervous hands)
• Final shot: reaction before action

Dialogue Pacing:

• Long, wandering conversations about nothing
• Sudden topic shifts that reveal danger
• Pauses that say more than words
• The mundane makes the violent more shocking

AI Application: For AI-generated dialogue scenes:

• Generate stable talking head footage
• Cut between speakers on power shifts, not line endings
• Use AI to extend pauses and create breathing room
• Generate insert shots of hands, objects, tells

---

Name

The Scorsese Track

Description

Use camera movement to express character psychology

When

Introducing characters or expressing their internal state

Example

SCORSESE'S MOVING CAMERA:

Camera Movement = Character Psychology:

TRACKING INTO: Confidence, power, entering a world

• "Copacabana shot" in Goodfellas
• Camera follows character who owns the space

PUSH IN: Increasing intensity, focus, obsession

• Camera approaches as emotion heightens
• Used in Mean Streets, Taxi Driver

SLOW MOTION + ROCK MUSIC: Mythologizing the moment

• Intro shots in Casino, Goodfellas
• Characters become legends in their own movie

FREEZE FRAME: This moment matters, remember it

• Often with narration
• Arrests time at key points

AI Application:

• Specify camera movement in prompts explicitly
• "camera tracks behind subject as they walk through [space]"
• "smooth steadicam following at eye level"
• Generate movement and static separately for more control

---

Name

Director's Shot List

Description

Create comprehensive coverage plan before shooting

When

Planning any video sequence for complete coverage

Example

PROFESSIONAL SHOT LIST TEMPLATE:

For each scene:

ESTABLISHING: □ Wide establishing shot (where are we?) □ Time-of-day indicator □ Atmosphere/mood shot

MASTERS: □ Wide master (full scene in one shot) □ Medium master (tighter, more energy)

COVERAGE: □ Over-the-shoulder A on B □ Over-the-shoulder B on A □ Clean single A (no B in frame) □ Clean single B □ Two-shot (both in frame)

INSERTS: □ Hands/action details □ Object inserts □ Reaction cutaways

MOVING SHOTS: □ Entry (character arrives) □ Exit (character leaves) □ Tracking (follow action)

SPECIAL: □ Point-of-view shots □ Subjective camera moments □ Any "signature" shots for this scene

For AI Generation: Prompt each shot type separately for maximum control. Match lighting/time across all shots. Generate masters first, use as reference for coverage.

---

Name

AI Cinematic Generation

Description

Generate film-quality footage using AI tools with director's intent

When

Using AI video tools to create cinematic content

Example

AI DIRECTOR'S WORKFLOW:

PHASE 1: VISION Before touching any AI tool:

• What emotion should the viewer feel?
• What's the visual reference? (film, director, era)
• What's the shot type? (establishing, close-up, tracking)
• What's happening in the frame?

PHASE 2: PROMPT ENGINEERING Structure prompts like a shot description:

"[SHOT TYPE] of [SUBJECT] [ACTION], [CAMERA BEHAVIOR], [LIGHTING STYLE], [VISUAL REFERENCE], [FILM QUALITY MARKERS]"

Example: "Close-up of woman's face, tears forming in eyes, camera slowly pushing in, warm golden hour backlighting, Spielberg-style emotional lighting, 35mm anamorphic, shallow depth of field, cinematic"

PHASE 3: GENERATION STRATEGY

• Generate multiple takes (3-5 variations)
• Separate foreground and background when possible
• Control pacing through generation length
• Use consistent seed/style across a sequence

PHASE 4: DIRECTOR'S SELECTION Evaluate each generation like dailies:

• Does it serve the emotion?
• Does it match the sequence's rhythm?
• Is the "performance" (movement, timing) right?
• Does it cut with adjacent shots?

Tools by Purpose:

• Veo3: Best for photorealistic humans, subtle emotion
• Runway Gen-3: Best for stylized, artistic looks
• Kling: Best for quick iterations, movement
• Sora: Best for complex physics, action

Anti-Patterns

---

Name

Coverage Without Vision

Description

Shooting every possible angle without a clear intent

Why

Creates editing confusion and wastes resources; the director's job is to know what the scene needs

Instead

Decide the emotional arc first. Shoot only what serves that arc.

---

Name

Cutting on Action

Description

Always cutting on movement because "that's the rule"

Why

Sometimes the tension is in the stillness; masters break this rule intentionally

Instead

Cut for emotion. Sometimes hold the uncomfortable pause. Let the actor finish.

---

Name

Unmotivated Camera

Description

Camera moves that don't express anything about the scene

Why

Every camera move should have a reason—revealing, following, intensifying

Instead

Ask "why does the camera move here?" If no answer, lock it off.

---

Name

Missing the Face

Description

Not capturing enough close-up reaction footage

Why

The face is where emotion lives; wide shots don't create empathy

Instead

When in doubt, get the face. Spielberg edits to faces. So should you.

---

Name

Flat Staging

Description

All actors at the same depth, evenly spaced

Why

Creates visual monotony; depth creates hierarchy and interest

Instead

Stage in depth. Foreground/background. High/low. Create visual layers.

---

Name

AI Autopilot

Description

Accepting AI generations without directorial judgment

Why

AI generates possibilities, not decisions; the director must curate

Instead

Review every generation against intent. Regenerate until right. Direct the AI.

Video Directing - Sharp Edges

Crossing The Line

Id

crossing-the-line

Summary

The 180-degree rule creates spatial confusion when broken unintentionally

Severity

high

Situation

When cutting between shots in a dialogue or action sequence

Why

The 180-degree rule maintains spatial consistency for viewers. An imaginary line connects the subjects—staying on one side keeps screen direction consistent. Cross it without intent, and characters suddenly seem to swap positions, confusing the audience about who's where.

Solution

1. Establish your axis (the line between subjects) 2. All cameras stay on one side of the line 3. If you MUST cross: use a neutral shot (straight on), a moving shot that crosses during movement, or cut to a close insert

For AI Generation:

• Specify camera position in prompts: "camera at 45 degrees left of subject"
• Generate matching angles: if one character faces right, other faces left
• When in doubt, generate frontal angles that work with any cut

Detection Pattern

Context

• dialogue
• conversation
• two-shot
• reverse
• over-the-shoulder

Intent

• cut
• edit
• sequence
• coverage

Jump Cut Jarring

Id

jump-cut-jarring

Summary

Jump cuts within the same angle are jarring without 30-degree rule

Severity

medium

Situation

When cutting between takes of the same subject/angle

Why

If you cut between two shots of similar angle (less than 30 degrees apart), the subject appears to "jump" in frame. This is jarring unless intentionally stylized (French New Wave, YouTube talking head style).

Solution

1. Change angle by at least 30 degrees between cuts 2. Or change shot size significantly (wide to close-up) 3. Or use insert/cutaway to bridge the cut 4. Or embrace it as style (match with music/energy)

For AI Generation:

• Generate 3 versions: wide, medium, close-up for each moment
• Use angle variations in prompts explicitly
• Generate B-roll specifically for covering jump cuts

Detection Pattern

Context

• same shot
• same angle
• talking head
• interview

Intent

• cut
• trim
• edit
• sequence

Eyeline Mismatch

Id

eyeline-mismatch

Summary

Mismatched eyelines break the illusion of subjects looking at each other

Severity

high

Situation

When cutting between characters in dialogue or reaction shots

Why

If Character A looks screen-right, Character B must look screen-left for them to appear to be looking at each other. Mismatched eyelines make characters seem to be looking past each other—breaking the scene's reality.

Solution

1. During shooting: mark eyeline targets on set 2. A looks right + up = B must look left + down (and vice versa) 3. For AI: specify eyeline direction in every prompt 4. Generate reaction shots with consistent eyeline to master

AI Prompt Template: "Subject looking [direction] at [height], matching eyeline for dialogue"

Detection Pattern

Context

• dialogue
• conversation
• reaction
• eye contact

Intent

• match
• cut
• reverse
• coverage

Empty Frames

Id

empty-frames

Summary

AI video generation often creates too much empty space

Severity

medium

Situation

When generating AI video for cinematic use

Why

AI models often default to centered subjects with lots of empty space. Professional cinematography uses deliberate framing—rule of thirds, leading room, balanced headroom. Empty frames feel amateurish.

Solution

1. Specify framing in prompts: "rule of thirds", "looking room left" 2. Request specific headroom: "close-up with minimal headroom" 3. Post-process: crop and reframe generated footage 4. Use wider generation, crop to intention

Framing Terms for AI Prompts:

• "extreme close-up, face filling frame"
• "medium shot, rule of thirds positioning"
• "subject positioned left with looking room right"

Detection Pattern

Context

• AI generate
• create video
• footage
• shot

Intent

• composition
• framing
• cinematic

Pacing Uniformity

Id

pacing-uniformity

Summary

AI-generated clips often have uniform pacing that feels robotic

Severity

medium

Situation

When using AI video for sequences that need dynamic rhythm

Why

AI models generate clips at consistent pace. Real cinematography varies— slow moments build tension, fast cuts create energy, pauses let emotion land. Uniform pacing feels mechanical and loses the viewer.

Solution

1. Generate clips at varying lengths intentionally 2. Use speed ramping in post (slow-mo, fast-mo) 3. Cut AI footage to match music/emotional beats 4. Intercut with static moments for contrast

Pacing by Emotion:

• Tension building: longer shots, slow push-ins
• Action/excitement: shorter shots, faster cuts
• Emotional beats: hold on faces, let moments breathe
• Resolution: return to longer shots, calm rhythm

Detection Pattern

Context

• AI video
• generated footage
• sequence
• edit

Intent

• rhythm
• pacing
• tempo
• cut

Lighting Inconsistency

Id

lighting-inconsistency

Summary

AI-generated shots often have inconsistent lighting across a sequence

Severity

high

Situation

When generating multiple AI video shots for the same scene

Why

Each AI generation creates its own lighting conditions. Cut between them, and the light direction/color/intensity jumps—destroying the illusion that these shots exist in the same space and time.

Solution

1. Specify identical lighting in every prompt for a scene 2. Use reference images with consistent lighting 3. Prompt: "consistent with previous shot, same golden hour lighting" 4. Post-process: color grade to match across shots

Lighting Prompt Template: "[shot description], [specific light direction e.g., key light from left], [color temperature e.g., warm 5600K], [time of day e.g., golden hour], [style reference e.g., Roger Deakins style]"

Detection Pattern

Context

• AI generate
• multiple shots
• sequence
• scene

Intent

• lighting
• match
• consistent
• continuity

Temporal Inconsistency

Id

temporal-inconsistency

Summary

AI video can have physics glitches, temporal stutters, and impossible movements

Severity

high

Situation

When using AI video for realistic footage

Why

AI models don't understand physics—objects morph, gravity fails, movements skip frames. These artifacts immediately break immersion and signal "AI-generated" to viewers, even unconsciously.

Solution

1. Review every frame, not just beginning and end 2. Use shorter generations (more control, fewer artifacts) 3. Cut away before artifacts (use B-roll to hide them) 4. Regenerate problem sections, don't try to fix 5. Composite: use AI for backgrounds, real footage for critical action

Quality Checklist: □ Do hands/fingers render correctly throughout? □ Does motion flow without stutters? □ Do objects maintain consistent form? □ Does physics make sense (gravity, momentum)?

Detection Pattern

Context

• AI video
• generated
• footage
• realistic

Intent

• review
• quality
• check
• artifacts

Coverage Gaps

Id

coverage-gaps

Summary

Shooting without comprehensive coverage leaves editors with no options

Severity

high

Situation

When planning shots for a scene that will be edited

Why

Directors often shoot the "hero shots" they envision but miss the connective tissue—the insert shots, reactions, cutaways. In edit, you discover you can't cut around problems because you have no coverage.

Solution

1. Always shoot: wide master, medium, close-ups, inserts 2. Get "safety shots"—neutral footage that can bridge any cut 3. Shoot 3x more B-roll than you think you need 4. Get reaction shots even when dialogue is finished

Coverage Minimum per Scene: □ Wide master (whole scene in one shot) □ Medium two-shot □ Singles on each character □ Inserts (hands, objects, details) □ Reaction shots (nodding, listening, thinking) □ Entry and exit shots

Detection Pattern

Context

• shot list
• coverage
• scene
• shooting

Intent

• plan
• shoot
• capture
• film

Actor Direction Vague

Id

actor-direction-vague

Summary

Vague direction to performers leads to generic, uncompelling performances

Severity

medium

Situation

When directing talent (or prompting AI digital humans)

Why

Telling an actor "be sad" or "look happy" gives them nothing to work with. Great directors give specific, playable actions—what does the character WANT in this moment? What are they DOING to get it?

Solution

1. Direct actions, not emotions: "Try to make them laugh" vs "be happy" 2. Give obstacles: "You need to tell them, but you don't want to hurt them" 3. Use sense memory: "Remember the last time you lost something precious" 4. Be specific: "On this word, you realize she knows"

For AI/Digital Humans:

• Prompt with specific micro-expressions
• Describe the thought behind the look, not just the emotion
• "Subject suppressing a smile, trying to stay serious"
• "Eyes searching for hope but expecting disappointment"

Detection Pattern

Context

• perform
• act
• expression
• direction
• digital human

Intent

• direct
• instruct
• emotional
• react

Scope Before Story

Id

scope-before-story

Summary

Prioritizing visual spectacle over emotional story

Severity

medium

Situation

When planning ambitious visual sequences

Why

It's tempting—especially with AI tools—to lead with "cool shots" rather than story necessity. But spectacle without emotional stakes is hollow. Audiences don't remember the shot—they remember how it made them feel.

Solution

1. For every "big shot," ask: what emotion does this serve? 2. Earn the spectacle with story setup 3. James Cameron's rule: the most expensive shot must be the most emotional 4. Use spectacle to reveal character, not replace it

Story Questions Before Any Shot:

• What does the audience need to feel?
• What does the character want?
• How does this shot advance that?
• Would the story work without this shot?

Detection Pattern

Context

• epic
• spectacle
• ambitious
• scale
• visual

Intent

• plan
• design
• create
• shoot

Ai Over Reliance

Id

ai-over-reliance

Summary

Letting AI make creative decisions that should be the director's

Severity

high

Situation

When using AI video generation without clear directorial intent

Why

AI generates possibilities—but it doesn't understand story, emotion, or what the sequence needs. Using AI without vision creates generic content. The director must know what they want BEFORE asking AI to create it.

Solution

1. Write the shot list before touching AI tools 2. Describe each shot's purpose in one sentence 3. Reject AI generations that don't serve the vision 4. AI is a tool—you're still the director

Before Any AI Generation: □ What emotion should viewer feel? □ Where does this shot sit in the sequence? □ What comes before and after? □ What specific visual serves this moment? □ What should I reject, no matter how "cool" it looks?

Detection Pattern

Context

• AI generate
• create
• footage
• video

Intent

• make
• create
• generate
• produce

Forgetting Audio

Id

forgetting-audio

Summary

Treating audio as secondary to visuals in planning

Severity

high

Situation

When planning visual sequences without audio consideration

Why

Directors obsess over shots but forget that audio carries half the experience. A mediocre shot with perfect audio beats a beautiful shot with bad audio. Sound design, music, and silence are directing tools as powerful as the camera.

Solution

1. Plan audio alongside visuals from pre-production 2. Know where music hits, where silence matters 3. Consider practical sound in every location 4. For AI: generate audio references before visuals

Audio Director Questions:

• What should the audience HEAR in this moment?
• Is there music? Does it start/stop/shift?
• What's the ambient sound of this world?
• Where is deliberate silence powerful?

Detection Pattern

Context

• scene
• sequence
• shot
• plan
• storyboard

Intent

• plan
• design
• create
• shoot

Video Directing - Validations

Shot List Coverage Check

Id

shot-list-completeness

Description

Validates that shot list includes essential coverage types

Check Type

content

Severity

warning

Pattern

Shot list should include:

• Establishing/wide shots
• Master shots for each scene
• Singles/close-ups on key subjects
• Reaction shots
• Insert/detail shots
• Entry and exit coverage

Fix

Add missing coverage types to shot list: □ WIDE: Establishing shot showing full location □ MASTER: Full scene coverage in continuous shot □ SINGLES: Individual character shots □ CU: Close-ups for emotional beats □ INSERTS: Detail shots (hands, objects) □ REACTIONS: Cutaway faces for editing flexibility

Emotional Intent Declaration

Id

emotional-intent-defined

Description

Each scene should have clearly defined emotional intent

Check Type

content

Severity

warning

Pattern

Scene/sequence descriptions should include:

• What emotion the audience should feel
• What the character wants in the scene
• The emotional arc (where it starts/ends)

Fix

Add emotional context to scene planning: "SCENE: [description] EMOTION: Audience should feel [specific emotion] CHARACTER WANT: [character] wants to [specific goal] ARC: Starts [emotion A], ends [emotion B]"

Eyeline Direction Check

Id

eyeline-specification

Description

Dialogue scenes should specify eyeline directions

Check Type

content

Severity

warning

Pattern

Dialogue coverage should specify:

• Character A eyeline direction (left/right, up/down)
• Character B eyeline direction (opposite of A)
• Any POV shots match character heights

Fix

For each character in dialogue:

• A looks: screen-[direction] at [height]
• B looks: screen-[opposite direction] at [height]

Example: "A looks screen-right, eye level. B looks screen-left, slight up."

Lighting Specification Check

Id

lighting-consistency-spec

Description

Multi-shot sequences should specify consistent lighting

Check Type

content

Severity

error

Pattern

For AI generation prompts in a sequence:

• Light direction (key light from left/right/above)
• Color temperature (warm/cool/neutral, or Kelvin)
• Time of day (dawn, golden hour, midday, dusk, night)
• Lighting style reference (cinematographer name or film)

Fix

Add to every prompt in a sequence: "[shot description], key light from [direction], [color temperature] lighting, [time of day], [reference cinematographer/film] style"

Camera Movement Motivation Check

Id

camera-movement-motivation

Description

Moving camera shots should have stated purpose

Check Type

content

Severity

info

Pattern

Camera movement specifications should include:

• What type of move (track, dolly, crane, steadicam, handheld)
• Why this movement (reveal, follow, intensify, disorient)
• What it expresses about the scene/character

Fix

For each moving shot: "MOVE: [type] - [direction] PURPOSE: To [reveal/follow/intensify/express] EXPRESSES: [what this communicates emotionally]"

AI Prompt Cinematography Elements

Id

ai-prompt-cinematography

Description

AI video prompts should include cinematographic specifications

Check Type

content

Severity

warning

Pattern

AI video prompts should specify:

• Lens/focal length feel (wide, normal, telephoto)
• Shot size (extreme CU, CU, MCU, MS, MLS, LS, ELS)
• Camera angle (eye level, low angle, high angle, Dutch)
• Film stock/format reference (35mm, IMAX, anamorphic)

Fix

AI Prompt Template: "[Shot size] of [subject] [action], [lens feel e.g., 85mm portrait, 24mm wide], [angle e.g., low angle looking up], shot on [format e.g., 35mm anamorphic, ARRI Alexa], [lighting], cinematic, shallow depth of field"

Sequence Rhythm Planning

Id

sequence-rhythm-plan

Description

Sequences should have defined pacing rhythm

Check Type

content

Severity

info

Pattern

Action/montage sequences should specify:

• Overall tempo (frenetic, measured, building, declining)
• Shot length pattern (long-short alternation, accelerating cuts)
• Music/beat alignment points

Fix

Add rhythm plan to sequences: "RHYTHM: [overall tempo] PATTERN: [e.g., 4s-2s-1s-1s accelerating to climax] BEATS: Music hits at [moments], cut on [beats] BREATHERS: Pause at [moments] to let emotion land"

Edit Flexibility Coverage

Id

coverage-for-edit

Description

Coverage plan should include edit flexibility shots

Check Type

content

Severity

warning

Pattern

Coverage should include "escape" shots:

• Neutral angles for any cut point
• Reaction shots without dialogue
• Insert cutaways
• Environmental/atmosphere shots

Fix

Add to every scene's shot list: "EDIT FLEXIBILITY: □ Neutral wide (can cut to from any shot) □ Listening reactions (no dialogue, just face) □ Hands/objects (cover any moment) □ Environment (buy time, establish rhythm)"

Performance Direction Specificity

Id

actor-direction-specificity

Description

Performer direction should be specific and playable

Check Type

content

Severity

info

Pattern

Direction should avoid:

• Vague emotions ("be sad", "look happy")
• Result-oriented direction ("be more compelling")

Direction should include:

• Playable actions ("try to make them laugh")
• Specific intentions ("hide that you know")

Fix

Convert vague direction to specific: VAGUE: "Look nervous" SPECIFIC: "You know something's wrong but you're trying not to let them see it. Every time they look at you, look away casually."

VAGUE: "Be angry" SPECIFIC: "You're not showing the anger—you're controlling it. The calmer you seem, the more dangerous you are."

Story-First Spectacle Check

Id

story-serves-spectacle

Description

Spectacle/effect shots should serve story purpose

Check Type

content

Severity

warning

Pattern

Visual spectacle moments should be justified by:

• Emotional story purpose
• Character revelation
• Narrative payoff

Not just: "It would look cool"

Fix

For each major visual moment, document: "SPECTACLE: [description] STORY PURPOSE: This reveals/advances [what] EMOTIONAL GOAL: Audience should feel [emotion] EARNED BY: Setup in [previous scene/moment]"

Audio Integration Planning

Id

audio-integration-plan

Description

Visual plans should include audio considerations

Check Type

content

Severity

warning

Pattern

Scene planning should include:

• Music placement (where it starts/stops/shifts)
• Sound design notes (ambient, effects, silence)
• Dialogue priority (clean recording needs)

Fix

Add audio layer to scene planning: "AUDIO:

• Music: [starts at/ends at/shifts at]
• Ambient: [environmental sound]
• Effects: [specific sound design needs]
• Silence: [deliberate quiet moments]
• Dialogue: [recording priority areas]"

Axis/Line Awareness Check

Id

180-rule-awareness

Description

Coverage plans should identify and maintain spatial axis

Check Type

content

Severity

error

Pattern

Dialogue/action scenes should specify:

• The line (axis between subjects)
• Which side camera stays on
• Any planned line crosses (with technique)

Fix

Add to dialogue scene planning: "AXIS: Line between [A] and [B] CAMERA SIDE: All coverage from [left/right] side LINE CROSS: [None / At [moment] via [technique]]

If crossing: Use neutral shot, dolly through, or character movement to justify"