Make a Shot List for the Proof-of-Concept Scene You Have Chosen

Most shot lists fail by addition. A scene ends up with fourteen entries because each one looked defensible in isolation, and somewhere around the ninth sits a slow tilt along a door handle that explains nothing and interrupts everything.

For a proof of concept the failure is sharper. The list covers the actions and forgets the reactions, so the exact thing the scene was built to test never appears on screen. The scene shoots cleanly, cuts together, and answers no question.

The example below is invented for this lesson. Nobody has shot this scene, no edit exists, and no audience has watched it. What follows is a paper method: build a small list, find the shot that isn't earning its place, and replace it rather than adding to it.

Restate the proof question before you list a single shot

Say back to yourself what this scene was selected to find out. Not what it's about. Not how it could look. The uncertainty.

Take a scene with two people at a doorway: a resident who lives inside and a visitor who arrives. The resident holds out an envelope. The visitor refuses it. The resident sets the envelope down and withdraws. The visitor takes it.

The uncertainty this scene exists to test is whether a change of mind reads with no dialogue. Can a viewer watch someone decline something, then watch them accept it, and understand that a decision happened in between — rather than reading the second moment as an unrelated grab at an unguarded object?

Four things have to land for that to work:

1. The visitor refused the envelope deliberately.
2. The resident left it and withdrew.
3. The visitor decided otherwise.
4. The visitor took it.

Only the retrieval is an event that carries its own meaning. The other three depend on a reading: the refusal must look deliberate rather than inattentive; the deposit and withdrawal is an observable action whose meaning must land as a concession that is not an agreement; the decision must look like a change of mind rather than an opportunistic grab. A list that shoots the retrieval and skips what has to be read covers one thing out of four.

That's the standard every entry has to meet. Not would this look good — that question belongs to a showreel — but does the viewer come out of this shot with something the scene needs them to have. If you can't name it, you've found the candidate for removal.

Break the scene into actions, reactions, and necessary transitions

Write the beats before you write the shots. Keep the list short enough that you can hold it next to the shot list and check them against each other.

• B1. The visitor stands outside the threshold. The resident stands inside it. The door is open.
• B2. The resident offers the envelope.
• B3. The visitor refuses.
• B4. The resident sets the envelope down on the visitor's side of the threshold and withdraws.
• B5. Nothing yet.
• B6. The visitor takes the envelope.

B2 through B4 are actions. B3 is also a reaction — the visitor responding to something the resident did — and it's the one where the reading can go wrong, because a refusal that plays as inattention isn't a refusal. B2→B3 is a response. B3→B4 is a concession that is not an agreement: the resident doesn't accept the refusal, they simply leave the object anyway. B4→B6 is the only transition in the scene with nothing inside it, and that gap is where the whole proof lives.

Two spatial facts have to be pinned down, because shots depend on them.

The deposit surface is on the visitor's side of the threshold. If the resident sets the envelope on a table inside and pulls the door shut, then B6 requires the visitor to knock, open the door, or break in — three different scenes, none of them the one you chose. The surface's location isn't set dressing; it's the reason the last beat is possible at all.

The door's resting state is a story condition. How far the resident withdraws decides what kind of act the visitor's decision is. If the resident steps two paces back and the door stands wide open, the visitor is not alone, and changing their mind becomes defiance in front of a witness. If the door comes nearly shut, the visitor is effectively unobserved, and the decision is a private one. Different performance, different read, different shot. Decide it now; cover it later.

Reactions are the part of a beat list people skip, because actions are easier to picture than responses. An offer is a picture. A refusal is a picture. The half-second where someone changes their mind is not a picture until you decide what it looks like.

Give each planned view an identifiable purpose

Write the purpose as an information claim, not a subject. "Close-up of visitor" is a subject. "After this shot the viewer knows the envelope is available and the resident has gone" is a claim, and you can tell whether it's true.

Here is one entry in full, as an example of the fields — not as a standard format:

S04 — B4 — deposit and withdrawal View: medium wide, holding the deposit surface and enough of the door to show how far the resident goes. What the viewer gains: the object is now available, and the resident has removed themselves. Dependency: the door's resting state. The reaction beat's read depends on it, so this shot establishes a condition, not just a beat.

Five columns are enough: a stable reference, the beat, the view, the gain, and the dependency. What matters isn't the form. It's that the gain column exists, that it can be blank, and that you check it. A blank gain is a shot looking for a reason.

S01 and S06 are the required bookends. S02 supplies the offer the refusal answers. S03 carries the refusal as deliberate rather than inattentive. S04 establishes that the envelope is available and the resident has withdrawn. S07 carries the change of mind. Those four shots do distinct work, and S07 is the one that was missing. S05 is texture.

Texture is legitimate. A production with room for it should buy some. This list has no room, because it has one job: protect a single uncertain read from being crowded out.

Keep the identifiers stable. Notice that S07 doesn't become "S05." Retiring a number and reissuing it to a different shot is how a list quietly corrupts itself: someone still holding the previous version reads "S05" and arrives expecting a door handle. The new shot gets a new number. That's why story order now reads S01, S02, S03, S04, S07, S06 — it looks wrong for a second, and then it clarifies that an ID is a name, not a position. The retired row stays visible in the table because deleting it destroys the cross-reference for anyone with an older copy.

Read the planned cut and repair missing coverage

Now walk the list in story order and ask a single question at every cut: where does the viewer learn what they need for the next action?

The original list: S01, S02, S03, S04, S05, S06.

The viewer sees where everyone stands. They see the offer. They see the refusal. They see the resident leave the envelope and withdraw. Then they see a door handle. Then the visitor has the envelope.

The reader can build a coherent story out of that — they took it once nobody was watching — and that story is not the one the scene was built to test. Worse, the handle insert isn't sitting harmlessly to one side. It occupies the exact slot in the cut where the decision could have been. The scene spends its one available gap on a door.

The revised list: S01, S02, S03, S04, S07, S06.

Same six shots. The viewer sees the refusal, the withdrawal, a person alone with a decision, and the retrieval. The change of mind now has a location.

The temptation is to keep S05 and add a seventh shot. Run both arrangements and watch what happens. Put the reaction after the insert and you've placed a door handle between the trigger and the response, delaying the one beat that matters. Put the insert after the retrieval and it just costs a setup for a picture doing nothing. Either way the list grew by one and the evidence didn't.

Replace before you add. Before any new entry goes on the list, look for one that can leave. Here the removal and the addition were the same decision; usually they aren't, but the question should always be asked, because a producer reads the list as an ask.

Don't over-correct. If the reaction is performed as a person visibly wrestling with temptation, the change reads — and you've answered a different question. The uncertainty was whether restraint reads. A theatrical beat proves that theatre reads, which nobody doubted. Plan the smallest version that keeps the decision visible, and leave the beat's duration marked as unsettled in the list. An honest note on an uncertain entry is worth more than a settled-looking list that hides one.

What this pass does and doesn't tell you. A paper walkthrough shows whether every required reading has a home. It cannot show whether the reading lands, because that takes shooting it, cutting it, and watching it with someone who wasn't on set. It also can't tell you whether the scene is affordable, safe, or shootable in the time you have. Comparing these two cuts is a comparison of plans, not an audience test.

Prepare for production discussion without claiming feasibility

Your list now has two orders, and they are not the same order.

Story order is what the viewer sees: S01, S02, S03, S04, S07, S06. This is the version a cinematographer and an editor need, because it shows the intended cut on paper in the order it will play.

Setup order is the day. Grouping shots by lighting direction, camera position or lens is normal and sensible — the publisher's own guide to making shot lists describes entering shots as attributed records and regrouping them for production setups (the guide was checked on 2026-09-18; its embedded videos and the application itself were not tested). Treat that as one provider's way of holding a list, not a format anyone enforces, and not as evidence that this scene works.

Regrouping must not move a dependency. You can shoot S06 first if the light is right; the envelope still has to be on the step and the door still has to be in its decided resting position before the take. A setup order is not an edit order, and nobody should read it as one.

The list also can't answer the questions people will ask it to answer. How long a setup takes, who's needed on the day, whether a rig is safe, what any of it costs — those belong to the first assistant director, the key grip, the cinematographer and whoever holds the budget. Ask them, and hand them the editorial-order list so the conversation is about coverage rather than about paper.

What the list can do is expose dependencies while there's still time to change a plan. A cinematographer can tell you whether one frame can carry two pieces of information at once. An editor can tell you whether a beat has enough room to breathe. Ask both before the shoot, when their answers are still cheap.

You're finished with this document when every view on it can name what the viewer gains, and every dependency you haven't resolved is written down where a collaborator can find it.

The list ends here at S01, S02, S03, S04, S07, S06 — one shot removed, one added, no more shots than before. The change of mind still isn't proven, and no paper pass will prove it. But the scene is no longer being asked to produce a reading it has nowhere to put, and the missing reaction has somewhere to happen. That's what a proof-of-concept shot list is for: not to guarantee the answer, but to make sure the question gets asked in front of a camera.

Frequently asked questions

What determines whether a shot belongs on a proof-of-concept list?

The proof question, meaning the uncertainty the scene exists to test. Each view should name what the viewer gains for a later beat. If you cannot name that gain, the shot is a candidate for removal; attractiveness alone belongs to a showreel.

Why list beats before shots?

To see actions, reactions, and necessary transitions. Actions are easier to picture than responses, so reactions and the gap where a change of mind happens are often skipped. If the scene tests a reaction, that reaction needs a location.

What is wrong with fixing a missing reaction by adding a shot?

Replacement should come before addition. In the example, a door-handle insert occupies the exact slot where the decision could go; keeping it and adding a reaction delays or spends a setup without improving the evidence. The revised list keeps the same six shots by removing one and adding one.

What can a paper walkthrough prove?

It can show whether every required reading has a planned home and can expose dependencies while plans are still changeable. It cannot show whether the reading lands, or whether the scene is affordable, safe, or shootable; those require production experts and eventually shooting, cutting, and watching with someone who was not on set.

Why keep shot IDs stable even if story order changes?

An ID is a name, not a position. Renumbering corrupts cross-references, so someone holding an older version may arrive expecting a different shot. The retired row stays visible in the table because deleting it destroys that reference.

Further reading

• Trailer, Sizzle, or Proof of Concept: Choose the Moving Material for a TV Pitch — Read this to separate a test from a trailer: fragments hide the difficult part the test exists to include, and the material's label sets what it can claim.
• Use Your Own Location Photos to Test a Film Idea, Not Pretend It Is Shot — When a spatial dependency is unresolved, such as which side of the threshold the surface sits on, this shows how to answer it with your own photographs.