Skill: Filmmaking & Cinematography
What
Professional film techniques — shot types, camera movements, lighting setups, composition rules, and the 180-degree rule — translated into AI image generation prompt language. Think like a DP (Director of Photography), not an illustrator.
Why
AI models trained on billions of images have internalized cinematic conventions. When you use film language precisely, the model produces images that feel like movie stills — proper framing, motivated lighting, professional composition. Vague descriptions produce vague images. Specific cinematic direction produces cinematic output.
1. Shot Types — The Complete Reference
By Subject Size
| Shot | What's visible | When to use | Prompt language |
|---|
| Extreme Wide (EWS) | Landscape dominates, figures tiny | Establishing location, isolation, scale | "Extreme wide shot — [figure] tiny against [vast environment], emphasizing [scale/isolation]" |
| Wide / Establishing (WS) | Full environment + full figures | Opening a new scene, showing spatial relationships | "Wide establishing shot of [environment]. [Characters] visible full-body, [spatial relationship]" |
| Full Shot (FS) | Full body, head to toe | Character introduction, showing body language/costume | "Full shot of [character] from head to toe, [pose], [environment context]" |
| Medium Wide (MWS) | Knees up (aka "cowboy shot") | Action + character, showing what hands are doing | "Medium wide shot, [character] visible from knees up, [action with hands/body]" |
| Medium (MS) | Waist up | Standard dialogue, most versatile | "Medium shot, [character] from waist up, [expression], [hand gesture]" |
| Medium Close-Up (MCU) | Chest up | Intimate conversation, showing emotion + some gesture | "Medium close-up, [character] from chest up, [detailed expression], [shoulder/hand detail]" |
| Close-Up (CU) | Face fills frame | Emotion, reaction, critical dialogue | "Close-up on [character]'s face. [Precise expression detail]. [Lighting on face]" |
| Extreme Close-Up (ECU) | Single feature (eyes, hands, object) | Tension, detail reveal, symbolic moment | "Extreme close-up on [specific feature]. [Granular detail]. [What this reveals]" |
By Angle
| Angle | Camera position | Psychological effect | Prompt language |
|---|
| Eye Level | Camera at subject's eye height | Neutral, equal, documentary | "Eye-level shot, camera at [character]'s height" |
| Low Angle | Camera below, looking up | Power, dominance, intimidation | "Low angle looking up at [character], [they appear] towering/powerful/imposing" |
| High Angle | Camera above, looking down | Vulnerability, weakness, overview | "High angle looking down at [character], [they appear] small/vulnerable/exposed" |
| Bird's Eye | Directly overhead | Omniscience, pattern, detachment | "Directly overhead view, looking straight down at [scene/pattern]" |
| Worm's Eye | Ground level, extreme low | Dramatic power, alien perspective | "Worm's eye view from ground level, looking up at [subject] against [ceiling/sky]" |
| Dutch Angle | Camera tilted on its axis | Unease, disorientation, tension | "Dutch angle, frame tilted 15-20 degrees, creating visual unease" |
| Over-the-Shoulder (OTS) | Camera behind one character, looking at another | Dialogue, connection, POV-adjacent | "Over [Character A]'s shoulder (back of head and shoulder visible), facing [Character B]" |
By Movement (Translated to Stills)
Camera movements become implied in single frames:
| Movement | Film technique | Still image equivalent | Prompt language |
|---|
| Dolly In | Camera physically moves closer | Tighter framing than previous panel | "Closer now — [subject] larger in frame, background compressed" |
| Dolly Out | Camera physically pulls back | Wider framing, more environment visible | "Pulled back — more of the room visible, [subject] smaller in context" |
| Tracking/Dolly | Camera moves alongside subject | Subject stays same size, background slides | "Moving alongside [character], background shifting as we keep pace" |
| Crane Up | Camera rises vertically | Higher vantage, more overview | "Camera has risen — now looking slightly down, more floor visible, [subject] below" |
| Pan | Camera rotates left/right | Different part of same space visible | "Same position, but turned to face [new direction] — [new elements] now visible" |
| Tilt | Camera rotates up/down | Different vertical portion visible | "Same position, now looking [up/down] — [ceiling/floor/sky] dominates frame" |
| Whip Pan | Ultra-fast pan | Motion blur in direction of pan | "Horizontal motion blur across entire frame, suggesting rapid camera swing to [direction]" |
| Steadicam/Handheld | Slightly unstable, organic | Subtle imperfection in framing | "Slightly off-center framing, not perfectly composed — organic, lived-in feeling" |
2. The 180-Degree Rule
The Rule
In any scene with two subjects, draw an imaginary line between them. Keep the camera on ONE SIDE of that line for the entire scene. Crossing it reverses their positions on screen and disorients the viewer.
In Sequential AI Generation
Scene: Character A (left) talking to Character B (right)
CORRECT (all panels from same side):
Panel 1: A at left, B at right (wide shot)
Panel 2: A at left, B at right (medium on A)
Panel 3: A at left, B at right (medium on B)
WRONG (crossing the line):
Panel 1: A at left, B at right
Panel 2: A at RIGHT, B at LEFT ← viewer is confused
Prompt Implementation
Establish positions in the first panel and maintain them:
- Panel 1: "A stands at the LEFT side of frame, B stands at the RIGHT, facing each other across a desk"
- Panel 2: "Close on A (at LEFT of frame), turned right toward B"
- Panel 3: "Close on B (at RIGHT of frame), turned left toward A"
When to Break the Rule
Only deliberately, for disorientation:
- A character switches allegiance
- Reality shifts (dream, flashback)
- Dramatic reveal that recontextualizes the scene
When breaking intentionally, add a neutral "bridge shot" (overhead or straight-on) between the sides.
3. Lighting Setups
Three-Point Lighting (The Foundation)
| Light | Purpose | Prompt language |
|---|
| Key Light | Main illumination, creates form | "Primary light from [direction], casting [shadow description]" |
| Fill Light | Softens shadows from key | "Softer secondary light from opposite side, filling shadows to [level]" |
| Back/Rim Light | Separates subject from background | "Edge of light along [shoulder/hair], separating figure from background" |
Lighting Ratios for Mood
| Mood | Key:Fill ratio | Prompt shorthand |
|---|
| Happy/bright | 1:1 to 2:1 | "Evenly lit, soft shadows, bright and open" |
| Neutral/natural | 3:1 | "Natural lighting, visible but soft shadows" |
| Dramatic | 4:1 to 8:1 | "Strong directional light, deep shadows on opposite side" |
| Noir/tense | 8:1+ | "Harsh single-source light, nearly black shadows, dramatic contrast" |
Named Lighting Setups
| Setup | Description | Best for | Prompt |
|---|
| Rembrandt | Key at 45° creates triangle of light on shadow cheek | Dramatic portraits | "Rembrandt lighting — key light at 45 degrees above and to the right, creating a triangle of light on the shadow-side cheek" |
| Butterfly | Key directly above and in front, shadow under nose | Glamour, beauty | "Butterfly lighting — light directly above, small shadow under nose, even cheekbone highlights" |
| Split | Light hits exactly half the face | Duality, conflict, secrets | "Split lighting — half the face in bright light, half in deep shadow, line of division down the center" |
| Practical | Light from sources visible in frame (lamps, screens, windows) | Naturalism, realism | "Lit only by [the desk lamp / phone screen / window], [warm/cool] light, everything else in shadow" |
| Silhouette | Subject backlit, front completely dark | Mystery, anonymity, drama | "Subject in complete silhouette against [bright background], no facial detail, outline only" |
Light Quality
| Quality | Character | Prompt language |
|---|
| Hard light | Sharp shadows, defined edges, dramatic | "Hard direct light, crisp shadow edges, no diffusion" |
| Soft light | Gradual shadows, gentle transitions, friendly | "Soft diffused light, gentle shadow gradients, no harsh edges" |
| Dappled | Broken light through leaves/blinds | "Dappled light filtering through [source], patterns of light and shadow across [surface]" |
| Volumetric | Visible light rays (god rays, dust motes) | "Volumetric light rays visible cutting through [dusty air / fog / steam]" |
4. Composition Rules
Rule of Thirds
Divide the frame into a 3x3 grid. Place subjects at intersection points, not dead center.
- "Subject positioned at the right-third intersection, not centered"
- "Horizon line along the lower third, sky dominating upper two-thirds"
Leading Lines
Lines in the scene that guide the eye toward the subject.
- "Corridor walls creating converging lines that draw the eye toward the figure at the far end"
- "Rows of desks creating diagonal lines pointing toward the teacher at the front"
Frame Within a Frame
Use elements in the scene to create a secondary frame around the subject.
- "Seen through the doorway — the doorframe creates a frame-within-frame around the figure in the room"
- "Visible through a gap in the crowd, framed by shoulders on either side"
Negative Space
Empty space that emphasizes the subject or creates mood.
- "Figure small at the bottom-left, vast empty sky filling the upper-right — emphasizing isolation"
- "Large expanse of empty wall behind her, subject pushed to the edge of frame"
Symmetry (Deliberate)
Perfect symmetry creates formality, power, unease (Kubrick, Wes Anderson).
- "Perfectly symmetrical composition — hallway centered, identical doors on each side, figure dead center"
- "Symmetrical framing, centered on the throne, columns matching on both sides"
Foreground Interest
Place something in the near foreground to create depth and draw the viewer in.
- "Blurred plant leaves in the near foreground frame the conversation behind them"
- "Edge of a coffee cup at the bottom of frame, out of focus, the meeting visible beyond it"
5. Scene Coverage — The Standard Pattern
For any dialogue scene, filmmakers shoot a predictable set of angles:
1. MASTER/ESTABLISHING — wide shot, shows full environment and all characters
2. MEDIUM TWO-SHOT — both characters visible, waist up
3. OVER-THE-SHOULDER A → B — behind Character A, seeing B's face
4. OVER-THE-SHOULDER B → A — behind Character B, seeing A's face
5. CLOSE-UP A — reaction, emotion
6. CLOSE-UP B — reaction, emotion
7. INSERT/CUTAWAY — detail shot (hands, object, clock)
You don't need all 7 for every scene. Pick from this menu based on what the scene needs:
- Simple exchange: 1 + 3 + 4
- Emotional scene: 1 + 5 + 6
- Tense scene: 3 + 4 + 7 (tight, claustrophobic, no breathing room)
6. Color Grading as Storytelling
Color Temperature Moods
| Temperature | K range | Mood | Prompt |
|---|
| Warm golden | 3000-4000K | Nostalgia, comfort, memory | "Warm golden light, like late afternoon sun, amber tones" |
| Neutral daylight | 5500K | Reality, present moment, documentary | "Neutral daylight, true colors, no color cast" |
| Cool blue | 7000-10000K | Clinical, cold, isolation, modern | "Cool blue-white light, fluorescent feel, slightly sterile" |
Complementary Color Palettes
| Palette | Emotional register | Prompt |
|---|
| Teal + Orange | Modern cinema, action, drama | "Teal shadows and warm orange highlights, modern cinematic color grade" |
| Blue + Yellow | Hope amidst despair, dawn/dusk | "Blue ambient shadows with pockets of warm yellow light" |
| Red + Green | Nature, Christmas, but also tension | "Deep greens with accents of red — vibrant but slightly unsettling" |
| Desaturated + one color | Focus, symbolic emphasis | "Nearly monochrome scene with only [object color] saturated — it draws the eye" |
7. Lens Choice (Prompt Translation)
You can't specify actual focal lengths to AI models, but you can describe the visual characteristics:
| Lens feel | Visual effect | Prompt language |
|---|
| Wide (24-35mm) | Exaggerated depth, spacious rooms, slight edge distortion | "Wide-angle view, room feels spacious, slight perspective exaggeration, near objects larger than life" |
| Normal (50mm) | How the eye sees, natural proportions | "Natural perspective, no distortion, how the eye naturally sees the scene" |
| Telephoto (85-200mm) | Compressed depth, foreground/background squeezed together | "Compressed perspective, background appears closer than it is, shallow depth of field, intimate feeling" |
| Macro | Extreme close-up on tiny details | "Macro view, extreme detail on [tiny subject], shallow focus, rest of world dissolved" |
8. Emotion-to-Technique Quick Reference
| Desired Emotion | Shot Type | Angle | Lighting | Composition |
|---|
| Power | Low angle, full shot | Worm's eye | Rim light, low-key | Symmetry, centered |
| Vulnerability | High angle, close-up | High angle | Soft, high-key | Negative space, off-center |
| Tension | ECU, Dutch angle | Low or canted | Split, low-key | Tight framing, no breathing room |
| Intimacy | CU, MCU | Eye level | Soft, warm, golden hour | Rule of thirds, shallow DOF |
| Isolation | EWS, wide shot | Any | Blue hour, cold | Negative space, tiny centered figure |
| Menace | Low angle, Dutch | Below eye level | Under-lighting, split | Shadows dominating |
| Joy | Medium, wide | Eye level | High-key, golden hour | Open framing, bright, warm |
| Dread | Slow push-in sequence | Eye level or low | Low-key, practical | Leading lines toward threat |
| Revelation | Dolly out sequence | Any | Dark shifting to lit | Frames-within-frames opening up |
| Chaos | Handheld, Dutch | Variable | Practical, flickering | Broken composition, motion blur |
| Mystery | Silhouette, OTS | Various | Backlit, rim only | Foreground obstruction, shadows |
9. Sequential Frame Patterns
Pre-built multi-panel sequences for common storytelling moments. Use these as starting templates.
The Approach (4 frames)
- EWS — character tiny, destination visible in distance
- WS — character closer, destination more defined
- MS — character's expression visible, determination or fear
- CU — character's face at the moment of arrival
The Reveal (3 frames)
- Reaction shot — character's face seeing something (shock, awe, horror)
- POV or insert — what they're seeing
- Wide — the full context, character AND the revealed thing together
The Confrontation (5 frames)
- Establishing two-shot — both characters, spatial relationship clear
- OTS on Character A — speaking, asserting
- OTS on Character B — reacting, challenging
- CU on A — emotional peak, the key line
- CU on B — the response that changes everything
The Power Shift (4 frames)
- Eye-level two-shot — characters equal
- Low angle on Character A — A appears dominant
- High angle on Character B — B appears diminished
- ECU on B's face — defeat, submission, or defiant fury
The Discovery (3 frames)
- MS — character searching, exploring, moving through space
- CU — character's expression changes (recognition, shock, understanding)
- Insert shot — the discovered object/clue/evidence in sharp detail
Common Mistakes
- Using film jargon the model doesn't understand — "dolly zoom", "whip pan", "rack focus" don't translate. Describe the VISUAL RESULT instead.
- Centering everything — default composition. Push subjects off-center using rule of thirds.
- Forgetting the 180-degree rule — characters swap sides between panels, disorienting the reader.
- Unmotivated lighting — dramatic shadows with no visible light source. Always specify WHERE the light comes from.
- Same shot size every panel — monotonous. Vary between wide/medium/close to create rhythm.
- Describing camera movement instead of the result — don't say "dolly in", say "closer now, subject larger in frame."
- No establishing shot — jumping to close-ups without first showing the space. Start wide, then go tight.
- Ignoring color temperature — mixing warm and cool light without motivation. Each light source has a temperature.