Cinematic Shot Direction

Direct cinematic shots as editorial decisions

Treat a shot as a controlled change in audience knowledge, attention, and feeling—not as a bundle of cinematic adjectives. Decide what the audience must perceive, choose the camera relationship that makes that perception likely, and specify only the variables needed to preserve it through production and editing.

Use three evidence labels throughout the work:

• Established fact: optics, geometry, or supported perceptual evidence.
• Production convention: a broadly used continuity or craft practice that may be broken deliberately.
• Production heuristic: a defensible starting point, not a law or guaranteed emotional effect.

Do not claim that a high angle always means weakness, a low angle always means power, a wide lens always feels energetic, or handheld always feels real. These are contextual readings. State the intended effect and test whether the complete image supports it.

Begin with the dramatic job

Write one sentence before choosing gear or movement:

> At this beat, the audience should notice [subject/change], understand [relationship/geography], and feel [quality], while [information] remains concealed or secondary.

Then identify the shot's editorial function:

• Orient: establish people, objects, exits, distances, or direction of travel.
• Reveal: introduce new information by cutting, reframing, focus, blocking, or camera travel.
• Witness: hold a behavior long enough for performance and consequence to register.
• Identify: align with a character's looking, hearing, or bodily position.
• Compare: place subjects in one frame or matched frames so their relationship is legible.
• Isolate: remove contextual information to privilege face, gesture, texture, or object.
• Transition: carry direction, shape, action, occlusion, or attention into another shot.

If the sentence still works with a locked camera, prefer the locked camera until movement has a specific job. A moving camera can add meaning, but the ASC's professional guidance begins movement choice with the script and intended shot, not the rig (ASC, “Tools for Camera Movement”).

Build the frame from relationships

Choose shot size by information, not prestige

Shot-size boundaries are flexible; they describe how much of the subject and environment is visible, not an absolute camera distance (Oklahoma State University, “Cinematography”). Use the smallest scale that still contains every required piece of information.

| Scale | Primary information | Frequent tradeoff | |---|---|---| | Extreme wide / establishing | geography, scale, destination, isolation | face and fine action become unreadable | | Wide / full | full-body action, blocking, spatial relationship | subtle expression loses weight | | Medium wide | body language plus meaningful environment | neither face nor geography dominates | | Medium | gesture, interaction, conversational relation | feet, entrances, and larger geography disappear | | Medium close-up | face with some posture and context | hand action and spatial relation may be lost | | Close-up | expression, decision, tactile detail | geography and partner response disappear | | Extreme close-up / insert | one decisive feature, object, or action | context must come from adjacent shots |

Name the framing landmark when precision matters: “full figure with feet and doorway,” “mid-torso,” “shoulders and head,” or “eyes and mouth,” rather than relying on abbreviations alone. For objects, name the visible surfaces and scale cue.

Compose for a priority order:

1. Primary subject or action.
2. Secondary relationship needed for meaning.
3. Background evidence needed for place or consequence.
4. Negative space reserved for gaze, motion, graphics, or an upcoming reveal.

Inspect edges, occlusions, mergers, horizons, and bright distractions. Aspect ratio is part of the shot contract: Kodak recommends explicitly communicating the intended frame because downstream framing otherwise becomes guesswork (Kodak, “Aspect Ratio Targets”). For vertical and horizontal deliveries, redesign blocking and negative space; do not assume a center crop preserves the shot.

Separate distance, height, and angle

These are independent controls:

• Distance determines the camera's spatial relation to the subject and is the primary control of perspective.
• Height is the lens center's elevation relative to the subject: floor, waist, chest, eye, overhead.
• Angle is where the optical axis points or where the camera sits around the subject: level, looking up/down, frontal, three-quarter, profile, rear three-quarter.
• Roll is horizon tilt. Keep level unless imbalance or a graphic diagonal is intentional.

Specify both height and angle. “Low angle” is ambiguous: a camera near the floor can remain level, or an eye-height camera can tilt upward. The first changes foreground and horizon placement; the second changes convergence and which surfaces are seen.

Treat camera position as point of view. Ask who is permitted to see from here, what the position reveals, and whether the audience is observing, participating, pursuing, trapped, or judging. Do not reduce this to a fixed symbolism dictionary.

Use lens and perspective correctly

Established facts: focal length and sensor/image format jointly determine angle of view; focal length also changes magnification. Perspective—the relative apparent size and spacing of objects—is governed by camera position. “Telephoto compression” emerges when a longer lens is paired with a more distant camera position to maintain subject size, not from focal length alone (Sony, “Focal Length, Angle of View & Perspective”; Nikon, “Understanding Focal Length”).

Choose in this order:

1. Place the camera for the desired perspective and relational distance.
2. Choose the angle of view that includes the needed frame.
3. Choose aperture/focus behavior for attention, exposure, and practical focus tolerance.
4. Evaluate optical character only after the geometry works.

State focal length as a format-aware field of view: “moderate wide, about 28 mm full-frame equivalent,” not merely “28 mm.” If the capture format is unknown, specify “wide / normal / long field of view” plus desired distance and framing.

Practical relationships:

• Move close with a wide field of view to enlarge near/far differences, show more environment, and make camera presence felt.
• Move back with a long field of view to flatten relative depth, simplify background coverage, and observe from a distance.
• Hold camera position and change focal length to crop or widen without changing perspective.
• Keep subject size but change focal length and distance together to change perspective.
• Treat shallow focus as an attention tool with continuity cost. Depth of field changes with camera distance, focal length, and aperture (Sony, “What is depth of field?”).

The ASC advises choosing lens, camera placement, and key light consciously as a system, and notes that a wide lens can support multiple actions in one frame while a longer lens can isolate a subject from farther away (ASC, “Fresh Perspectives”).

Choreograph movement and blocking together

First choose what moves:

• Subject only: the frame becomes a stage; blocking creates emphasis and depth.
• Camera only: the viewer's relationship changes while action holds.
• Subject and camera together: tracking can preserve size while backgrounds and relationships evolve.
• Neither: duration, performance, light, and sound carry the beat.

Name camera movement geometrically and dramatically:

| Move | Geometry | Useful dramatic job | |---|---|---| | Pan / tilt | rotate from a fixed position | redirect attention, connect offscreen space, follow action | | Dolly in/out | translate toward/away | cross emotional or informational distance; reveal parallax | | Truck/track | translate sideways | follow travel, expose layers, compare subjects | | Pedestal/boom/crane | translate vertically | reveal hierarchy or geography through changing height | | Arc/orbit | travel around subject | change background, alignment, or relational side | | Handheld | operator-borne micro-translation/rotation | embody instability, urgency, proximity, or human observation | | Stabilized follow | smooth free translation | accompany a subject through space without dolly rigidity | | Zoom | change focal length from fixed camera position | change framing/angle of view without translation or parallax | | Rack focus | change focus plane, not camera position | transfer attention between depth layers |

Specify a movement with start frame, path, motivation/cue, speed character, end frame, and what becomes newly visible. “Slow cinematic push-in” is incomplete. Prefer: “Start medium two-shot; when Mina hides the key, advance 60 cm on axis over four seconds, ending chest-up on Mina while Theo falls soft at frame edge; no pan.”

Block subjects with marks and beats:

• starting and ending positions;
• facing and eyeline target;
• path, pace, and action cue;
• foreground/background crossings;
• moments of occlusion or reveal;
• required separation and silhouette;
• relation to focus plane and light.

Rehearse camera, subjects, focus, and light as one system. A subject stepping toward the key may brighten and become frontal; crossing behind it may lose eye light; an arc can turn modeled side light into flat frontal light or silhouette. ARRI distinguishes key as the primary modeling source and fill as shadow control, while warning that hard opposing fill can create unnatural double shadows (ARRI Lighting Handbook). Track key side, softness, contrast, color, background level, practical positions, and motivated changes across coverage.

Preserve screen direction and editorial continuity

Define the active axis before coverage:

• two-person dialogue: line between the people;
• travel: direction of motion;
• look: line from observer to object;
• task: line connecting person, hands, and work surface.

Production convention: keeping cameras on one side of the axis preserves left/right placement, opposing eyelines, and direction of travel. Crossing may reverse those relationships (Columbia Film Language Glossary, “180-Degree Rule”; BBC, “Crossing the Line: Movement”). The convention is not a moral rule. Cross deliberately when reversal, rupture, or reorientation is the point, and bridge it with visible camera travel across the axis, a neutral/on-axis view, or a new establishing relation.

For each shot, record:

• who is screen-left and screen-right;
• gaze direction and exact eyeline target;
• entry and exit edge;
• motion direction;
• axis side and whether/where it changes;
• hand, prop, wardrobe, and action state at the cut;
• focus target and lighting state;
• start and end pose with handles.

Use match on action to carry attention and causal motion through a cut. Peer-reviewed evidence supports the idea that continuity techniques guide attention and reduce awareness of cuts; action coinciding with a cut is particularly effective (Smith & Henderson, 2008; Smith, 2012). An eyeline match links a look to its object (Columbia Film Language Glossary, “Eye-Line Match”). Preserve the causal order: look → object, action → consequence, question → reaction.

Design coverage around beats, not insurance

Coverage exists to construct the scene's intended rhythm and emphasis. Do not collect every subject at every size by default.

For each dramatic beat, determine:

1. Whose information or reaction controls the beat?
2. Must audience geography remain visible?
3. Is the decisive event simultaneous with another event?
4. Does performance need uninterrupted duration?
5. Where can an editor enter and leave the shot?

A robust minimal set often includes:

• one orienting or relationship view;
• one view for each decisive action or speaker;
• one reaction that changes interpretation;
• inserts only for story-critical details;
• an exit, transition, or re-establishing option;
• clean action overlap and handles.

The BBC's five-shot guidance—face close-up, action close-up, two-shot, over-shoulder, and an unusual “art” angle—is useful as a coverage check, not a mandatory formula; it also advises a meaningful change in both size and angle between adjacent shots (BBC, “Five Essential Shots”).

If a beat depends on simultaneity, preserve it in a two-shot, deep composition, split attention, or choreographed oner rather than fragmenting it into singles. If a performance depends on escalation, allow a sustained shot and reserve coverage for the actual turn. Coverage can live inside a moving master when blocking, focus, and exposure support it.

Write a generative production shot specification

Describe one continuous shot per specification unless the production system explicitly supports independently controllable multi-shot output. Keep story context outside the visual instructions when it does not affect what appears.

Use this order:

SHOT ID / EDITORIAL FUNCTION
Intent: what changes for the audience in this shot.
Duration / delivery: approximate usable duration, frame rate if important, aspect ratio and crop obligations.
Subject and setting: visible identities, objects, surfaces, time, weather, and spatial anchors.
Start frame: shot size, composition, subject placement, camera distance, height, angle, roll.
Lens / focus: format-aware field of view, focus target, depth behavior, intentional breathing or distortion only if relevant.
Blocking: start mark, action path, eyeline, gesture/prop state, end mark.
Camera movement: geometry, path, cue, speed profile, stabilization character, end frame.
Lighting interaction: motivated source direction, key side, contrast, practicals, atmosphere, and how movement changes light.
Continuity: axis side, screen direction, entry/exit, previous and next action states, match point.
Exclusions / tolerances: forbidden extra subjects, text, cuts, camera operations, or changes; acceptable variation.
Review target: the two or three conditions that make the take usable.

Prioritize a single dominant subject action and a single dominant camera idea. This is a production heuristic for controllability, not a provider limit. If two moves are essential, state their sequence and cue rather than listing them as simultaneous style words.

Diagnose failures from evidence

Review the actual frames or clip before rewriting direction. Identify the first frame where intention breaks.

| Symptom | Likely production cause | Repair | |---|---|---| | “Long-lens compression” is absent | focal length changed but camera position did not | move camera back, restore subject size with longer field of view, preserve background anchors | | Face is distorted or room stretches | camera too close for intended perspective | move back; select narrower field of view; restage foreground |

…

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: calesthio
• Source: calesthio/generative-media-skills
• License: MIT
• Homepage: https://github.com/calesthio/OpenMontage

Install and usage instructions live in the source repository linked above.