Build a Rough Cardboard Set Model for a Commercial Pitch
Build a Rough Cardboard Set Model for a Commercial Pitch
A treatment can say that a partition hides a doorway until the camera moves. It cannot say how far the camera has to move before the doorway appears, and that distance is usually the only part of the sentence a director and a director of photography will actually argue about. A rough cardboard model can produce that distance in an afternoon, from card, glue and a kitchen table.
What it cannot produce is authority. It is an exploratory sketch model, not the white-card production model that arrives later, not a survey of any real location, and not evidence about lenses, lighting or full-size camera access. David Neat's 2014 notes on white-card models for film and television design treat the finished production model as a distinct later stage from the quick exploratory one, and that distinction matters here: you are not building the thing that gets approved. A 2022 account of a white-card meeting at the Lyceum describes a small model, a scale figure and movable elements used to talk through a proposed set — theatre's version of the same conversation, not a guarantee that a miniature reproduces full-scale light, safety, cost or access.
So the aim here is narrow. One room, one question, one scale, two views, and a written record of the guesswork.
Put the question and dimensions on the base
Start with the sentence you cannot resolve, not with a drawing of the set. For this exercise the invented sentence is: from the entry position the partition hides the doorway; as the camera tracks right, part of the doorway appears. The model's job is to turn "part of the doorway appears" into a number.
Now put numbers on it. The plan below is invented for the exercise, and everything from here on is arithmetic worked on paper — no cutting, gluing, measuring or photographing has happened in this article. The build is your job.
Interior clear floor: 4.00 m wide × 3.00 m deep. Supplied by the director as an assumption about the space.
Ceiling height: 2.70 m. Estimated. Nobody has confirmed it.
Doorway, back wall: 0.90 m wide, 2.05 m high, near jamb 1.90 m from the left clear face. Consider it supplied; it is the thing the shot depends on.
Partition: 1.70 m long, 2.40 m high, 100 mm nominal thickness, standing 0.80 m in front of the back wall, spanning from 1.20 m to 2.90 m across the room. Proposed. You invented it, which means you can move it later — that is the point of a sketch model.
Camera positions: A at 0.50 m in from the entry face, 2.20 m from the left clear face. B at the same depth, 3.40 m from the left clear face. Eye height 1.60 m.
Choose a scale and keep it. 1:25 is a reasonable default for room-sized interiors: at 1:25, one meter becomes 40 mm. The 3.00 m depth becomes 120 mm. The 4.00 m width becomes 160 mm. Nothing else in the model gets to be at a different scale, including the figure you will add later.
One dimension is still missing, and it is the one that causes most of the trouble. A drawn plan line has no thickness. A built wall does. At 1:25, a wall that is 100 mm thick on site is 4 mm in your hand — which means the card you choose is a design decision, not a stationery decision.
Write all of this on the base in two colors. Supplied dimensions in one, estimates in another, with the estimates plainly labeled. Something like:
SKETCH MODEL — NOT A SURVEY
SCALE 1:25 (1 m = 40 mm)
INTERIOR CLEAR 4.00 × 3.00 m — ASSUMED
DOORWAY NEAR JAMB 1.90 m FROM LEFT CLEAR FACE — ASSUMED
CEILING 2.70 m — ESTIMATED, UNCONFIRMED
WALLS 100 mm NOMINAL — MODEL WALLS 4 mm (TWO LAMINATED SKINS)
PARTITION 0.80 m OFF BACK WALL — PROPOSED
A model photographed well enough for a treatment looks authoritative. The label is what stops a good photograph from erasing the assumptions underneath it.
Resolve the corners before cutting every wall
Measure your card before you trust its label. Stack ten sheets, measure the stack, divide by ten. If the stack comes to 20 mm, your stock is 2.0 mm, which reads as 50 mm at 1:25. That is thinner than the 100 mm wall the plan assumes, so laminate two skins for the walls the question depends on and get 4 mm.
Now build a corner before you build a room. This is the part people skip, and it is the part that pays.
Cut a base at 160 × 120 mm — the plan's clear dimensions used, wrongly, as though they were the outside size of the room. Stand a 160 mm wall inside the left edge and measure the clear distance from its inner face to the right edge of the base. You get 156 mm. At 1:25 that is 3.90 m, and the plan says 4.00 m. Add the opposite wall and the clear width drops to 152 mm, or 3.80 m. The same thing happens in the other direction.
You have just lost 200 mm in each direction without the blade slipping once. The error is 8 mm on the model, which is nearly a quarter of the width of a doorway drawn at 36 mm.
The correction is easy once you see it. Cut the base to the outside footprint: 4.00 m plus two 100 mm walls is 4.20 m, which is 168 mm, and 3.00 m plus two walls is 3.20 m, which is 128 mm. Then draw the interior clear rectangle 160 × 120 mm, inset 4 mm from every edge, and build to that line. The walls now sit on the base rather than inside the room, and the clear space is what the plan says it is.
Two smaller decisions belong here as well. First, at a corner one wall runs past and one wall stops between the other's ends. Pick one convention, mark it on the card before cutting, and keep it for the whole model; a mixture of the two makes every corner slightly different and the plan drifts. Second, if the walls stand on the base, measure wall heights from the floor surface, not from the table. Otherwise the base's own thickness quietly adds 50 mm to every wall.
Keep the failed corner. Put it next to the corrected one. It is the most useful object in this build, because it is the record of how tight the question was.
Build the openings and keep the assembly square
Transfer the doorway from the same plan and the same datum. Measured from the left clear face, the near jamb at 1.90 m marks at 76 mm and the far jamb at 2.80 m marks at 112 mm. The head at 2.05 m is 82 mm up from the floor surface. Write "BACK" and an arrow on the inside face so the wall cannot be glued in backwards.
Cut on a protected surface with a steel rule and a fresh blade, taking several light passes rather than one heavy one, with your fingers behind the line rather than in front of it. That is the ordinary habit, not a safety regime — this is craft card, not a workshop material.
Lay the cut back wall flat on the base before anything is glued and check that the jamb marks line up with where the plan puts them. An opening in the wrong place is invisible on a loose piece of card and obvious once it is a wall.
For the joins, a small right-angle guide — an engineer's square, a set square, or the corner of a stout box — held against the two pieces while the glue grabs will do more for the model than any amount of finishing later. Base card and PVA are the whole material list. You do not need specialist plastics, laminating film or the adhesives that professional model-making sources use for their white-card work; those belong to a different job with different requirements.
And keep the model plain. No paint, no printed texture, no little furniture, nothing described as dressing. A dressed model hides layout errors, and layout errors are most of what a sketch model exists to expose. If the doorway is 8 mm off, you want to see it, not cover it.
Leave access for the view you need
The room needs a ceiling for the interior view to read as a room, but it must not be glued down. Rest it on the wall tops, or hold it with two strips of tape that peel off.
For the eye-height view, the practical problem is that no camera fits inside a 160 × 120 mm room. The honest solution is to lift away the entry wall, because the entry wall sits behind the camera and is not in the sightline at all — removing it cannot change the answer. Then add a card apron in front of the base so the camera can stand where the plan says it stands. Position A is 0.50 m from the entry face, which is 20 mm; with the wall off, that becomes about 24 mm from the base edge. Write the apron's purpose on it. The rule is simple: what you removed to see the room must not change the room's plan, and the overhead view should still show all four walls.
Add a height marker rather than a person. At 1:25, 1.60 m of eye height is 64 mm. A marker of the right height, standing at the door, gives you a reference you can check against the wall and the partition. It is an orientation aid, not a measurement proof.
Then take two photographs of the same geometry. The first is overhead, with the ceiling off and the phone held level and parallel to the base, so the image reads as the plan. The second is from position A at eye height, with the camera square to the back wall. Note on the base, or in the file name, which wall is missing in each frame.
Change the relationship, not the evidence
Now run the comparison the model was built for. Take the deliberately misdimensioned corner, correct the wall lengths, and repeat both views from the same declared positions. The temptation at this point is enormous and entirely human: you nudge the camera two centimeters, or tilt slightly, or crop, because the corrected shot does not look as convincing as the wrong one. That is the moment the exercise is testing. Same positions, same lens, same height, or the comparison is worthless.
The arithmetic gives you something to check against. In the corrected model, from position B, the sightline grazing the partition's right end leaves about 135 mm of the doorway visible — 5.4 mm on the model, which is small enough that your marking has to be careful. Run the same calculation on the broken build, where the interior has shrunk to 3.80 × 2.80 m and the partition has been laid out from the base's back edge rather than from the wall's inner face, so its gap to the back wall reads 0.70 m instead of 0.80 m. You get roughly 119 mm instead. A difference of about 16 mm, produced entirely by wall thickness, with the design unchanged.
That is why the model's output is a margin, not a verdict. "135 mm of doorway appears" is a much more useful sentence in a treatment meeting than "the doorway is hidden" — it tells the director the reveal is real but thin, and it tells the designer that the partition's thickness and its distance from the back wall are now load-bearing decisions rather than details.
What you must not do is carry any of this past its limit. The model says nothing about what lens will produce that framing, whether the reveal reads on screen at all, or whether a real camera can make the move on a real floor with a real operator. It says nothing about lighting. It says nothing about whether the back wall is where you have assumed it is. Those questions go to the location assessment and to the people who own them, carrying the 135 mm with them as a number to confirm rather than a number to build from.
Stop there. The model has answered its declared question, and cutting anything else — a second partition, a doorway in the side wall, the beginnings of a corridor — only adds dimensions that nobody asked about and that the base is no longer labeled to support.
Photograph the corrected model from both positions, with the failed corner and the label visible in at least one frame, and the plan the same in both. What you hand over is not a beautiful object. It is a stated scale, a written list of assumptions, a documented error, and one number with a known margin around it — which is more than the treatment had before the card came out.
Frequently asked questions
What is the rough cardboard model for, and what is it not?
It is an exploratory sketch model that turns a shot question into a rough number, such as how far the camera must move before a doorway appears. It is not the later white-card production model, not a survey of a real location, and not evidence about lenses, lighting, or full-size camera access.
What should go on the base before cutting?
The unresolved sentence, the scale, and every dimension labeled as supplied, assumed, estimated, or proposed. A practical label might include interior clear dimensions, doorway position, ceiling height, wall thickness, and partition placement, so a good photograph cannot erase the assumptions underneath.
Why build a corner before building the whole room?
Because using the clear dimensions as the base's outside size quietly loses wall thickness in both directions. In the worked example, a 160 × 120 mm base yields 156 mm clear, then 152 mm once the opposite wall is added; cutting the base to the outside footprint and drawing the clear rectangle inset corrects it.
How should openings and assembly be kept accurate?
Transfer the doorway from the same plan and datum, mark which face is the back, lay the cut wall flat before gluing, and check jamb marks against the plan. Use a right-angle guide for joins, keep the model plain without dressing, and measure wall heights from the floor surface rather than the table.
What can the final comparison honestly claim?
It can give a margin under declared assumptions, such as roughly 135 mm of doorway visible from one position versus roughly 119 mm in the broken build, with the difference produced by wall thickness. It says nothing about lens choice, screen readability, lighting, real camera access, or whether the assumed back wall is where it is.