Chapter 5 · Shot design and camera language

Yasser Ashour

2 hours ago

28 min read

Part 1 · Before the camera

A shot is a decision: what the audience must see, from where, for how long, and why this shot and not another. On a set, a good crew fills the gaps you leave. A model fills them too, with whatever it has seen most often. This chapter teaches the craft that leaves no gaps: the job of each shot, the breakdown that fixes it, the camera move as a sentence, placement and lens, the geography of a scene, and what "bold" means on a high-budget commercial.

In this chapter

The job every shot must do, and the breakdown that fixes it in prose
A whole scene broken down: a bottled-tamarind spot, shot from the floor of a Cairo repair yard
The camera move as a verb, a mechanism, a destination and a hold; how to size a move; the locked camera
Families of move, postures, parallax, wipes and transitions
Placement and lens as emotional decisions; geography, the geometry note and format
What bold means, and how to choose where a scene is bold

Before you start. Chapter 2 explains how a model reads words; Chapters 3 and 4 give the concept and the treatment that this chapter turns into shots. Chapter 8 makes the shots a shot table. Chapter 9 writes a still, Chapter 14 a moving shot, and Chapter 15 directs the people inside the frame. Each tool's own dialect for camera words is in its section (the table in 5.10).

5.1 What a shot list decides, and who decides it

In top-tier commercial production a shot list is not a coverage checklist. It is the point where the treatment stops persuading and becomes something a crew can execute. The usual order: treatment and references; boards and scout discussions; blocking tests or an animatic (a rough cut of drawn or photographed frames timed to sound); a shot list the first assistant director can schedule; revision on the day. Stink's Andzej Gavriss calls the storyboard a bible, to be interpreted rather than worshipped. Effects-led or rhythm-dependent spots start with tight boards; performance-led work stays open. The pattern is controlled openness.

Authorship is shared. The director owns each shot's dramatic and visual purpose; the cinematographer tests the lens, the height, whether the move is feasible and what the light will cost; the first assistant director needs the list because the schedule and the budget follow the number and difficulty of set-ups. With a model as the camera, nothing fills a gap for you. A human operator told "wide of the yard" still chooses a height, a lens and a moment. A model told the same invents whatever you left out. The breakdown is where you make those choices, before a credit is spent.

Count beats, not coverage. Ask four questions of every scene: how many story beats are inside its running time; how much of each beat blocking can carry without a cut; how much one frame can show at once; how much editorial punctuation the concept has already promised. Editors' and directors' accounts put a premium 30-second spot at about 8 to 20 principal set-ups, a 60-second narrative at about 15 to 35, and a 90-second film from about 25 upward. Treat the figures as a sense of scale, not a target.

Every shot must do a job that the shots around it cannot do. Name the job before you name the size. "What is this for?" comes before "how wide is it?"

A shot may carry one primary and one secondary job, never more than two. The vocabulary is richer than "establishing, medium, close-up".

Family	Jobs a shot can do
Narrative and structure	hook; thesis; world orientation; exposition compressed; orientation reset; transition bridge; handoff beat; escalation; reversal; ending payoff; release
Character and emotion	character truth; emotional withholding; emotional reveal; vulnerability proof; status frame; status inversion; pressure frame; intimacy proof; isolation beat; resolve
Product and brand	product proof; product seduction; behaviour proof; value contrast; brand mnemonic (the image the audience keeps); icon build; ritual beat; utility proof; social proof; cultural detail; world texture; sensory detail
Comedy and surprise	comedy set-up; comedy beat; deadpan hold; absurdity reveal; contrast gag; timing spike
Editorial and rhythm	rhythm beat; pace break; impact punctuation; the source of a graphic match or a sound bridge; montage accelerant; breath beat

Three patterns recur in good work. The wide is rarely an establishing shot: it is a thesis, a payoff or a spectacle frame, as the crescendo wide of Channel 4's We're the Superhumans is the payoff of everything before it. The close-up is there for proof, whether of truth, touch, comedy, stress or product: The Long Wait withholds the boy's release until one last face, and that beat resolves the film because every earlier frame was rationed. Blocking is editorial structure in space: in ON's Matt On Set the products pass from person to person along Matt McCreary's path without the movement stopping, so the blocking carries the story a montage would otherwise have to.

5.2 Writing the breakdown

Describe the frame in prose; keep the label as metadata. "Medium shot of hero in hallway" is not a design decision. It says nothing about where the person sits in the frame, what the background does, what presses on the edges or how the shot resolves. Compare: "Hero occupies the left third of frame at hip height; the empty hallway behind him stretches deep and cold, with a practical lamp flickering mid-background. He does not look at camera. The doorframe traps him on both vertical edges." A cinematographer, a designer, an editor and a model can all build that. Labels such as CU or OTS may follow the prose as shorthand and never replace it.

A full breakdown line has more fields than most shot lists. The table gives each field and where its answer ends up. The reasons stay in the shot record, and what the frame must show reaches the prompt (Chapter 9 teaches the three places a decision can live).

Field	What to decide	Where it goes
Shot and beat	a short tag and the beat's name	shot table
Intention	the job, in one sentence; at most two jobs	shot record; its visible result goes in the prompt
Framing in prose	where the subject sits, headroom or empty space, background layers, foreground, horizon and verticals, and which edges breathe and which press	prompt, as positions (Chapter 7 for edges)
Lens and optical intent	the focal length, stop and depth of field you would choose	reason in the record; lens and stop in the prompt beside what they should show (5.7)
Height and distance	floor, knee, hip, chest, eye level, overhead; how far; invasive, companionable, observational or detached	prompt, in real units or in body and furniture units
Blocking and path	who enters from where, crosses to where, turns where, leaves	prompt: positions in a still, order in a clip
Motion posture, endpoint, duration	the posture (5.5), where the camera lands and how long it holds; if static, why	record; the prompt gets the move sentence (5.4) or "The camera is locked."; length in settings (Chapter 14)
First frame, last frame, transitions	what must be visible at the first frame and at the cut; how the shot comes in and goes out	the first frame is the keyframe (Chapter 9); the last is the landing and hold; transitions are the edit's job (Chapter 29)
Light and colour	what lights the scene from inside its world; dominant and secondary colour; the protected brand or product colour	prompt, by cause and as colours on named surfaces (Chapter 6)
Performance and sound	what the actor must not overplay; the sound, word or breath that gives the shot its entry	prompt, as a task and cues (Chapter 15); the edit (Chapters 21 and 25)
Generation risk and plan	low, medium or high; what you will do about it	shot table; it chooses the route (Chapter 13)
Must protect	the one thing this shot cannot lose without losing its reason to exist	shot record; checked on every return

If a field cannot be filled with something specific, the shot is not ready. That is the point of the form: it forces decisions before money is spent, which is when they are cheapest.

Numbers belong beside their results. A focal length, a stop and a camera height in centimetres are the director's working language, and at finished grade they belong in the still prompt, each beside the visible result it should give ("a 40mm lens at T/2.8, the background dissolving"). A number is a cue for the look, never physics and never the only carrier of a decision. The makers write it this way: Google's prompt advice gives "a low-angle shot with a shallow depth of field (f/1.8)" as its example of directing like a cinematographer, and OpenAI's own portrait prompt asks for a "medium close-up at eye level, using a 50mm lens" while telling writers to treat camera specifications "as cues for appearance, not a guarantee of exact physical simulation" (both read 30 Sep 2026). At test grade, words alone are enough: "a portrait lens from an arm and a half away". In a clip the move carries its size and the still carries the lens (5.4; Chapter 14).

Planning for what a model finds hard

The breakdown must name each shot's generation risk before production, not after. A low-risk shot has one subject, one action, a short duration, a clean silhouette, readable light, a manageable camera and stable space. High-risk shots involve people touching, hands doing fine work, lip sync through a turn of the head, fast changes of blocking, reflective materials, busy layers of foreground and background, or exact continuity across several angles. For those, simplify the ask or design camouflage: keep hands at the edge or behind the product, choose a back three-quarter angle, crop before the anatomy becomes the subject, or split the beat into two shorter shots. Budget attempts as well as money: a risk and a mitigation on every line tell you where the attempts will go, and Chapter 13 turns each risk into a choice of route.

The first frame of a social film

For work that lives first on TikTok, Reels or Shorts the opening is stricter than on television. TikTok's own creative guidance asks for the proposition in the first three seconds and the hook within the first six, in a vertical 9:16 frame kept inside the interface's safe zone (read 30 Sep 2026). So frame one must already be the film. A logo, an empty scenic wide or a fade-up wastes the most valuable second you have. Open on a face with an unusual emotion, the product already in use, an unfinished action, something visibly wrong, or a movement strong enough to stop a thumb, and put the first sound there too: a voice, a breath, an impact.

5.3 A scene broken down: the yard

This is scene 2 of 3 in a 45-second spot for a fictional bottled tamarind drink, in a car-repair yard in Shubra, Cairo, at four on an August afternoon. Karim, 27, lies on a creeper under an old white Fiat 128 taxi with a seized bolt; only his legs and one arm show. His grandfather Amm Fathy, 72, a retired tram driver, sits on a plastic crate at the edge of the shade with a cold bottle sweating in his lap. Karim's hand comes out from under the car, palm up, without looking: he wants the spanner. The spanner lies on the concrete, within Fathy's reach. Fathy looks at it, looks at the bottle, and lays the bottle in the open palm. The fingers close, stop and feel the cold. A drop of condensation runs across the oil in the palm and leaves a clean line. Karim rolls out from under the car into the sun with the bottle in his fist, blinking. Nobody says a word. He asked for a tool and was handed a reason to stop.

One idea governs the camera: it lies down with Karim. Every frame in the scene is shot from the height of a man on a creeper, about 25 centimetres off the concrete. The car's underside becomes a dark lid across the top of each frame, Fathy is seen from below, and a hand that would be a detail from eye level becomes a face. The camera does not move until Karim does. The joke is carried by the arm's line, the light and the sound of a cap, not by the cut.

The axis runs along the car's flank. The taxi stands broadside to the camera, nose to the right. Karim is under the near sill; his arm reaches towards the crate at the front wheel, so the line from his hand to Fathy's crate runs left to right across every frame. All camera positions sit on the yard side of that line, so the reach never reverses direction on screen.

The bottle is planted. It sits in Fathy's lap from the first frame, beaded with water and darkening his thigh. The payoff is that it was always there. An object that appears on a cut reads as an editing trick and, in generated work, as an inconsistency.

Figure 5.1 — The yard from above. The axis runs from Karim's hand to Fathy's crate and is never crossed; the reach travels across the frame, never towards the lens.

The coverage is the minimum that tells the story. Six shots, each doing a job no other does.

Shot	Its job	Framing, in words	Enters → leaves
1 The yard	thesis; fix positions, heat, the planted bottle and the spanner	25 cm off the concrete, wide lens: the taxi broadside, its underside a black band across the top third; Karim's legs at left, one forearm out at the sill; the crate and Fathy right of centre; a bucket soft in the lower left corner	already reaching → cut as Fathy's eyes drop to the spanner
2 The palm	the request; sets up the clean line	on the ground, close: the open palm fills the lower half, oil in every crease, heat shimmer behind	still → the fingers twitch once, impatient
3 The look	the comic beat; the decision	from below, 30 cm high: Fathy's face, the crate, hard sky above, the taxi's edge cutting the upper left	mid-glance at the spanner → the bottle rising through the lower third
4 The lay-down	behaviour proof: cold, delivered by an object	25 cm high, a metre from the palm: palm at lower left, the bottle's base entering from upper right, held by a soft forearm	palm open → fingers curled round the cold
5 The drop	sensory detail; the image to remember	macro, on the palm's lines: a bead runs across the oil and leaves a clean track	bead at the wrist → the track holds
6 The roll-out	release; the light event; the only move	from position 1's side: the creeper rolls out of the shade, his face into hard sun; the camera rises from the ground to seated height with him	wheels emerging from shade → hold on his squinting face, the bottle at his chest

The alternatives that were considered and rejected teach as much as the plan.

Considered	Why it was rejected
An eye-level master of a man's legs under a car	it plays as slapstick, and the hand, which is the scene's face, becomes small
A top-down pour of the drink	the category's default; the proof in this scene is contact and cold, not liquid
A slow push-in on the label or the first sip	it announces the product; the pack shot belongs to scene 3
Shot and reverse shot between Karim and Fathy	it makes the exchange turn-based; the arm's line should make it one act

The scene chose where to be bold. Its boldness is in height and in the frame's edges: the car's underside presses the top of every frame like a lid, and Fathy is a monument seen from the ground. The light adds a second channel, the hard sun that Karim rolls into at the end. Movement is deliberately refused until the last shot. That refusal is a decision, written down: no push-in, no rack focus, no handheld, no slow motion, no lens wide enough to distort a face.

Every line of the full plan also carries its continuity (the bottle's state, the oil on the palm), its neighbours, the length it plays against the length to generate, a risk and a fallback. Shot 4 is the high-risk line, a hand and a bottle in contact: the plan is that the bottle is laid on the palm, not gripped, with the fingers curling in the next shot's first frames, and the fallback is to cut on the base touching the palm and show the closing fingers in an insert. Its light is written as a cause: hard sun from camera left, the bottle backlit and beading bright, the concrete bouncing warm light up into the palm. The camera is locked.

5.4 The move

Write every move as a verb, a mechanism, a destination and a hold. A move without a body floats: "a dolly's rolling weight" or "a slider's damping as it settles" gives it one. A move without a destination searches in its last second. Not "slow cinematic push-in", but a sentence with all four parts. For the yard's last shot: "The camera rises from the ground on a pedestal's counterweighted column, from a hand's height above the concrete to a seated man's eye level, keeping Karim's face in the lower third as he sits up, and by the end it has settled on his face against the workshop wall, the bottle at his chest, and holds." A rig name describes behaviour, never a make.

Template: the move sentence (template; not run). Fill every bracket or cut the clause; a locked shot replaces the whole sentence with "The camera is locked."

The camera [verb] [speed in words], with [mechanism: a dolly's rolling weight / a slider's damping / a hand-operated head], from [start framing] toward [destination], and by [the end / a named beat] it has settled on [landing framing] and holds.

Motivate the move from inside the scene. On a set the scene is blocked on its feet with the actors first, and the camera path is fitted afterwards. The camera follows forces: a car's acceleration, the direction of a gesture, a product's fall. When a character stands, the camera rises with her; when a hand drops under a table, the lens tilts down after it. A move that no action motivates should look designed, a precise mechanical path, never an aimless float.

Choose verbs that say what the camera is doing to the subject. "Moves" says nothing. Stillness: hold, pin, witness, wait. Precision: reframe, settle, isolate. Travel: creep, glide, push, pull, flank, tail, lead. Reveal: unveil, peel, emerge, uncover. Pressure: invade, close, corner, bear down. Energy: snap, whip, surge, slam to land. Scale: rise, descend, tower, dwarf.

When the camera should not move, say so: "The camera is locked." A declared static is a full design, and everything that moves is then the world's. Meaning comes from composition, light and the blocking inside the frame, and a lock can build tension before a release, as in the yard. Write down why static is the strongest choice; if you cannot, the shot may simply lack an idea.

Never write a hedge; name the rig or declare the lock. "Almost still", "subtle drift" and "subtle handheld" drift unpredictably, and the phrase "from frame one" stays banned (Chapter 14). Something visible moves at once, or the fixed camera is declared and the world carries the motion.

Give a shot one dominant camera intention. Two unrelated moves compete and raise warping, so split them: not "slow push-in, then pan left to the door". A named compound device counts as one intention when you write its start framing, trigger, movement, landing and hold, and keep everything else still: "Track alongside her at her walking pace, level with her chest, and as she reaches the revolving door ease ahead by half a step and resolve to a three-quarter frame of her face, then hold." Chapter 14 teaches the devices.

Size a move by what it does to the frame. A camera moving straight towards a subject makes it larger in inverse proportion to the distance, so a move can be sized before it is generated.

The camera moves	The subject appears	It reads as
10 % of the distance closer (15 cm from 1.5 m)	about 11 % larger	a drift, barely felt
25 % closer	about a third larger	a deliberate push
50 % closer	twice the size	a strong push
25 % farther	about a fifth smaller	a pull

A zoom changes the field of view and leaves the layers of the frame in the same relation; a dolly changes the relation between near and far, which is why parallax proves a real move (5.5). To check what came back, compare the subject's height in the first and last frames: their ratio is the ratio of the distances. How much of this to write differs by tool: a Kling move takes one approximate distance or angle per moving system with its duration and the visible result, never stacked; Seedance and Wan take pacing words and whole-second time stamps (16.3, 17.3, 18.3). Chapter 14 teaches sizing and measuring in full.

Write a dolly zoom by what the viewer sees. The camera moves away while the view narrows, or in while it widens, so the face keeps its size and the background changes behind it. Say what stays still and where the move lands: "The corridor behind her compresses and seems to draw closer while she stays the same size and holds, her eyes lifting to the light."

Name where a fast move lands. "The camera whip pans from the man on the left of the café table to the man on the right" gives the move something to aim at. Where a tool documents time stamps, add one (17.3), but the landing is what must be named.

Slow it before you simplify it, and simplify it before you split it. A fast move on a wordmark, a small held object or a turning face invites melting: slow it, reduce it to one motion system, protect the shape by name, then split it into two shots.

Refuse the default moves. The slow push-in on the reaction, the rack focus, the golden rim, the flare on the reveal and handheld "for intimacy" are what every generated advertisement does. Each can be right; earn it.

5.5 Families of move, and motion postures

Name the family before you write the move. Each family changes something different and fails differently.

Family	What changes	Write	Watch for
Orbit	the viewpoint round the subject	"circles a quarter of the way round her"	fragile on Kling: keep it simple and slow, or split it (16.3)
Track, dolly	the camera's position	"tracks alongside at her walking pace"	the most obedient family; parallax proves it; a zoom is not a dolly
Crane, rise, descend	the camera's height, on an arm	"crane down slowly, the heavy lowering of a crane arm"	one mechanism, named and put first, or it floats
Pan, tilt	the aim, from a fixed point	"pans right from the door to the window and stops"	no parallax

Advance and retreat are the most obedient family. A crane needs its arm, its weight and its strain, and only one body: a move that names a fire-truck ladder at the start and a crane arm at the end has two mechanisms and will float between them.

Parallax is the signature of a real move. The expensive frame is built in depth: a foreground anchor close to the lens (a column, wet leaves, textured glass), the subject in the middle ground, a background that gives scale. When the camera travels sideways the layers slide past at different speeds and the eye reads real space. A zoom enlarges everything equally; a pan swings everything together. If you want the audience to feel the camera travel, give it layers to travel past.

Figure 5.2 — Parallax. In a sideways move the layers separate by speed and the viewer reads depth.

Choose a posture, not an intensity. Directors and operators do not work with "moderate movement"; they work with named postures, each tied to what it does to the audience.

Posture	What it is	Use it when
Tableau lock	a dead lock; the world moves inside the frame	authority comes from stillness
Measured reframe	a pan, tilt or recompose from a fixed point	attention changes what it looks at
Controlled travel	a slow push, pull or slide	inevitability, pressure, attention intensifying
Companion track	tracking with, leading, following or flanking the subject	the camera is a partner, a witness or a co-runner
Reveal geometry	an arc, an orbit, a move round a corner or past a foreground	new information should emerge from space
Vertical authority	a crane rise, a jib drop, an overhead dip	a change of scale, lift or domination
Kinetic punctuation	a whip to a landing, a crash zoom, a snap tilt, a burst of handheld	a comic shock, an editorial shove, an impact

Common objectives map to moves and lenses as starting points. To reveal scale: a crane rising while it tilts down, on a wide lens, the subject in the lower third. To tighten tension: a slow push-in on a medium-long lens, right when earned. To convey isolation: a slow pull-out that lets the space grow round a centred subject. To follow speed: a parallel track on a normal lens with dense foreground to make parallax.

On a set, boldness climbs with the rig; with a model, risk climbs instead. Camera work ranks from precise statics, through single-axis slides and compound moves, to high-velocity sweeps and robotic motion control. A model removes the cost of the rig but not the difficulty of the image: two unrelated moves in one shot raise warping. Use complexity where it carries the idea.

Frame rate is part of the look. The award-winning narrative commercials keep classic cadence, 24 or 25 frames a second; a whole film at 60 tends to read as a showroom demonstration. What each tool returns is in its Settings section.

5.6 Point of view, wipes, transitions and a grammar that changes

Confine a point-of-view device to the one shot that is the point of view, and lock it: "no cuts, no zoom, natural head movement". Let the camera make a foreground wipe: an object crossing close to the lens and covering the frame for a moment. Name the object, its side and its direction, and keep the extras where they are.

Plan transitions in prep, not in post. Three are designed in advance on high-budget work. The block-and-cover: the camera tracks sideways behind a pillar, a door or a wall corner, the frame goes briefly dark or blurred, and the camera emerges somewhere else. The graphic match: two shots share a shape, a line or a direction of movement. The sound bridge: the next scene's sound arrives before its picture (a J-cut) or the last scene's sound runs over the next picture (an L-cut). In the yard spot the yard's radio runs under every cut so that six shots sit in one place. Chapter 29 cuts them.

Let the camera's grammar change when the story changes. In Coinbase's Your Way Out, the Film Craft Grand Prix at Cannes in 2026, a figure trapped in a video-game world begins in a fixed isometric camera; as he breaks out the camera becomes handheld and the set turns from game textures into real plants, steam and hard light. The grammar shift is the argument, and it gives a test for camera work: describe the camera in three words for the first third and the last third. If the words are the same while the story changed, the camera is weak. Weak is every shot a "beauty" shot, lenses chosen at random, the camera always moving; acceptable is consistent, competent coverage that means nothing in particular; aspirational is the yard's locked floor-level camera rising with Karim in its last shot.

5.7 Placement and lens

Write camera placement as a specification, and set each number beside the visible result it should give: height, distance, lens and stop, the subject's share of the frame, the angle and which side of the axis. Height and distance go in real units or in body and furniture units; either fixes a camera. Focal lengths in this book are full-frame equivalents. At test grade, words alone are enough: "camera beside the laptop lid at his eye level, about an arm and a half away, portrait lens feel". Chapter 9 puts the specification into element 8 of the twelve.

Written for this book · Food and drink · Nano Banana Pro, 16:9, text only · passage: elements 5 and 8 (frame inventory and camera placement) of the yard master, shot 1 · not run

Example 5.1 — a camera and a frame inventory as a specification

The camera sits 25 centimetres above the concrete, level, six metres from the crate, with a 28mm lens at T/4, so the whole yard stays sharp from about three metres to the far wall and near objects fall soft. In the foreground an overturned steel bucket cuts the lower left corner as a soft dark shape. In the midground a white Fiat 128 taxi stands broadside across the frame, nose to the right; its underside is a black band along the top third, Karim's legs show from the knee at the left, and one bare forearm reaches out of the shadow at the sill, palm up and open. At the right, in front of the taxi's front wheel, Fathy sits on a sun-faded plastic crate at the edge of the shade, a dark amber bottle upright in his lap, beaded with water; the frame's right edge cuts the crate's corner. A steel 14mm spanner lies on the concrete at the centre, an arm's length from the crate. In the background the low workshop wall runs out of focus.

Why it works

The height, distance, lens and stop are in real units, each beside the result it gives: "sharp from about three metres to the far wall, near objects soft".
The three planes are named, and so is what cuts each edge: the bucket at the lower left, the crate's corner at the right.
Nothing is described as moving: the arm is "palm up and open", the bottle "upright in his lap". The reach is a position.
The low height is the idea of the shot: from 25 centimetres the taxi's underside becomes a black lid across the top third.

Height is an attitude. Give height with precision (floor-skimming, knee, hip, chest, eye level, above the head, top-down) and give the feeling of the distance: invasive, companionable, observational or detached. A low camera elevates the subject without a word; Rina Yang's rule of thumb on Super. Human. was "If in doubt, go low". Generated images often look synthetic because they drift to dramatic angles for no reason. Choose the height; do not inherit it.

Name each shot's lens class, and what the lens must not do.

Class	Write (test grade)	What you get
Portrait	"a portrait-class lens from an arm and a half away"	the face fills without stretching; the background softens
Normal	"normal lens feel"	the room at human scale
Wide	"a wide lens from close, the room stretching away"	near objects grow; never near faces in singles
Long	"a long lens from across the room"	flat, stacked space; geography is lost

At finished grade the class becomes a number beside its result, as in Example 5.1. For the shot record, lens postures are easier to decide than focal lengths: wide and close (about 18 to 28 mm) when the camera must feel inside the action or the space is the concept; normal and frontal (about 35 to 50 mm) for human command and a product read in use; long and compressed (about 65 to 100 mm) for status, glamour and objects of desire, never the default for a whole project. Ask of every scene: which focal band dominates; what is lost if you go longer; what is lost if you go wider; which lens behaviour is forbidden. "Cinematic", "intimate" and "epic" are outcomes, not reasons.

A lens is an emotional decision. On Super. Human. the sport sequences used subtly anamorphic lenses to feel slicker, and the everyday scenes used spherical lenses, handheld and close, to feel grounded: the lens change is the change between myth and ordinary life. In Burberry's Open Spaces Justin Brown used 14 and 16 mm lenses so the camera could stay with performers in flight.

Anamorphic or spherical is a question of the lift the film needs, not of prestige. Anamorphic lenses leave oval out-of-focus highlights, horizontal flares and a particular compression of depth, and suit myth and a premium sheen that still feels physical. Spherical lenses suit handheld closeness, clean edges in tight spaces and legible products. Decide in the shot record; the prompt gets the lens class and what it does. Whether a tool renders an anamorphic character from words is not yet known.

Keep camera-body and lens-brand names out of every prompt. They work as style stickers, not physics: a camera name brings a colour-science and bokeh habit instead of a decision about distance, height and what the lens is forbidden to do. The same goes for a cinematographer's or director's name used as a style. The one narrow exception is a director's name that stands for a technique a camera can perform ("a Hitchcock zoom") in a Seedance 2.5 motion prompt, never as a visual style and never on Kling or Nano Banana Pro (17.3).

Lock the camera vocabulary to the world's period (a 1966 world moves on dollies, cranes and hand-operated heads, never on a stabiliser glide), and never ask for two incompatible spaces at once: "a close push and a full-body view" is a documented trigger of warping.

Read a reference's lens before you borrow it. A vast background with converging verticals means a wide lens; a background that looks large and stacked right behind the subject means a long one. Chapter 6 reads a reference for its light.

5.8 Geography and format

Fix the geography in the first second, and name the axis once. Sides exist only from the camera. An anchor object recurs in every view so the audience can orient itself, and a reverse angle comes from a plate made for that angle (Chapter 10), not from a guess.

Keep one geometry note per scene, and reuse it in every shot. It holds the landmarks, where each person is relative to them, which side of the axis the camera keeps, and who is in frame at the start. Write it in that order: place, landmark, relative positions, then screen placement. On a reverse, world positions stay and screen positions change.

The geometry note for the yard (no characters, no action, only the place from the camera). From the camera, the white taxi stands broadside across the yard, nose to the right; the low workshop wall runs behind it, with a painted advertisement fading at the far left; the yard gate is at the far left, where the sun comes in; the plastic crate stands in front of the taxi's front wheel, right of centre; the spanner lies on the concrete at the centre, an arm's length from the crate; an overturned bucket sits at the near left. Sides exist only from the camera; the camera stays on the yard side of the taxi.

Put the aspect ratio and the resolution in the settings, never in the prompt. When a tool does not offer the ratio the film needs (some offer no 1.85:1 or 2.39:1), make the nearest wider frame and crop or matte it in post; each tool's Settings section lists what it offers (11.6, 16.6).

Plan vertical as its own frame for any shot with a speaking mouth or a travelling object. A 9:16 frame is a different composition, not a crop of 16:9, for those shots. In the yard spot the vertical version reframes the palm and the lay-down tighter, opens on Fathy's face from below, and lets the bottle travel down the frame towards the palm rather than across it.

Format is a creative choice, not an administrative one. The Square Meter is square because its idea is square. Open Spaces chose 3:2 because the frame had to hold bodies rising. Super. Human. stayed in 16:9 for broadcast legibility and marked its two worlds inside the frame. 16:9 is the most product-lucid; a very wide ratio signals breadth but can work against full-body action with a product; 4:3 or square can create pressure, nostalgia or a graphic identity. Choose the ratio that is inseparable from the idea.

5.9 What bold means on a high-budget commercial

In the world of Cannes Film Craft, which rewards on-screen artistry that elevates or dramatically enhances an idea, bold is not busy. It is a posture: the camera declares an idea instead of recording a performance. Across the Film Craft winners from 2019 to 2024 the recurring pattern was not more movement but stronger authorship per shot: wider lenses used closer than drama usually allows, lower camera heights, more full-body blocking, cleaner silhouettes. Movement nearly always had a destination.

Film	Where it is bold
Hornbach, The Square Meter (Grand Prix 2024; Daniel Landin)	constraint: a square frame and a low, childlike height turn every one-square-metre room into a pressure box; the route through the rooms is the concept
Burberry, Open Spaces (Gold 2022; Justin Brown)	the camera does not admire flight from below; it flies with the bodies, on wide lenses, in a 3:2 frame
Channel 4, Super. Human. (Gold 2022; Rina Yang, directed by Bradford Young)	a split lens grammar: myth for the sport, documentary closeness for ordinary life
Kendrick Lamar, We Cry Together (Grand Prix 2023; Adam Newport-Berra)	one sustained, punishing formal idea with no release: a single take in which the performers' paths are the script
The New York Times, The Truth Is Worth It (Grand Prix 2019)	restraint: type, sound and pace do the work, and framing becomes information architecture

A commercial must declare. A commercial frame often has to establish a character, a tone, the brand's position and sometimes the product's use almost at once, so its lens, height, movement and blocking read more assertively than in art cinema, which may withhold. A commercial usually cannot, unless withholding is the idea.

Choose where the scene is bold. A scene cannot be moderate in every channel. Decide which channel carries its boldness (lens, frame geometry, camera height, movement, blocking or light) and make it strong in at least one, often two. "Beautiful but timid" is what you get when all six are only tasteful.

Figure 5.3 — Where two award-winning films and the yard scene put their boldness. Choose the channel scene by scene.

Restraint is bold only when it is singular. A still, quiet, atmospheric scene is valid, but on its own it drifts towards a prestige-drama mood board. It needs a counterweight: a wide lens close, a forceful low height, a strict axis, a product-aware composition, a transition. The Truth Is Worth It and We Cry Together win with restraint because their formal idea is intense enough to replace conventional boldness.

Give every scene its memorable shot. Instead of "a locked master and tasteful inserts", use this coverage model for high-budget work: open with a declaration, not a set-up; include one relation shot that makes the spatial or emotional problem legible; one blocking shot in which bodies or objects write the scene; one motion event where the camera's verb is strongest; one hero frame where subject and brand logic meet; and one shot whose only job is to be remembered six hours later. Read the yard this way: its declaration is the taxi's black lid across the master; its relation shot is the arm's line to the crate; its blocking shot is the lay-down; its one motion event is the rise; its hero frame is the palm with the bottle; its image to remember is the clean line the drop leaves in the oil.

What translates to a model, and what still breaks. Generated work handles designed movement without physical friction well: impossible transitions, architectural morphs, a camera path through a concept. What breaks fastest is what commercial cinematographers are paid to control: exact blocking continuity, repeatable eyelines, believable hands on objects, geography that stays consistent from lens to lens across cuts, and skin, cloth and hair under close scrutiny. Camera language therefore matters more with a model, not less. The 2026 juries made the same point from the other side, giving the Film Craft Grand Prix to a film made largely in camera.

5.10 Which tool for which job

Each tool reads camera words in its own dialect. This chapter gives the craft; the sections below give the words.

The job	Where to look
Placement, lens, height and frame inventory in a still	11.3 (Nano Banana Pro); 12.3 (GPT Image 2.5)
A locked camera, a measured slow move, a crane or a track from a start frame	16.3 (Kling); 17.3 (Seedance 2.5); 18.3 (Wan 3.0)
A named device (dolly zoom, whip pan, speed ramp)	17.3, and Chapter 14 for how to write the device
A move on Veo 3.1 or MiniMax H3	19.3
Which aspect ratios and resolutions a tool offers	the Settings section of each tool: 11.6, 12.6, 16.6, 17.6, 18.6
Sizing and checking a move; endpoints; start and end frames; format and length	Chapter 14
Who is in the frame and what they are doing	Chapter 15

5.11 The checks, the gate and the records

Before you sign a breakdown

Every shot has a job in one sentence, and no shot repeats another's job.
Every frame is described in prose, and every move has a verb, a mechanism, a destination and a hold, or the shot says "The camera is locked." and why.
One dominant camera intention per shot; any device has all five parts.
At finished grade, height, distance, lens and stop are in the prompt, each beside the result it should give; the reasons are in the shot record.
Every slow move is sized by what it does to the frame; a Kling move carries one approximate magnitude per moving system, a duration and a speed profile.
Every motion prompt ends on a composition you can cut on, then a hold.
The scene has a geometry note, and every shot keeps the camera on one side of the axis.
Every object that matters is planted before it pays off.
Generation risk and its plan are written for every shot; the riskiest have a fallback.
The scene has chosen where it is bold, and it has a memorable shot.
Vertical versions of speaking and travelling shots are planned as their own frames.

The breakdown passes the lock gate (Chapter 8) when every box is ticked; nothing is spent on a shot whose line is incomplete. This stage writes four things into the records: the shot table (one row per shot: job, framing, height and lens, move or lock, first and last frame, light state at entry and exit, risk and plan, must-protect); the geometry note and the plan drawn from above, one per scene; the camera relation for the scene (where it is bold, what it refuses); and, for each shot, the shot record's reasons behind every number that reaches the prompt.

Open questions

How each tool reads a camera height or a distance in real units is not yet known; treat every number as a cue and check the returned frame (Chapter 9).
Seedance 2.5 and Wan 3.0 do not document distances; their route is pacing words, ordered beats and whole-second stamps (17.3, 18.3).
The dolly zoom and other compound devices are unconfirmed across the tools; test one on a cheap frame before it carries a scene.

What to remember

Every shot does a job no other shot does; name the job before the size, and describe the frame in prose.
Put every decision in the breakdown first; a finished prompt carries what the frame must show, each number beside the result it should give.
Name the generation risk and the plan for every shot before you spend anything.
Write a move as a verb, a mechanism, a destination and a hold; write "The camera is locked." when it should not move; never write a hedge.
One dominant camera intention per shot. Size a slow move by what it does to the frame, name where a fast move lands, end on a composition and a hold, and refuse the default moves.
Height, distance and lens class are emotional decisions; keep brand names and people's names out of prompts.
Fix the geography in the first second, keep one geometry note per scene, and originate vertical frames for speech and travel.
Bold is not busy. Choose the channel where each scene is bold; a declared refusal counts, an absence of ideas does not.