A subject and a story fit in the same number of words, which is why almost nobody notices they have only got one of them. A film about a man who repaints the seat numbers in a stadium is a subject. It will get somebody through a shoot and leave them with material that will not assemble, and the thing it is missing can be written down and checked before anything is recorded.
The piece here is two minutes long and invented for this page. Ivo unlocks the east stand of a football stadium in Rijeka at six in the morning, in August, and repaints the numbers on the seats before the season starts. He has done it for thirty years.
Two sentences, and whether they differ
The test is mechanical. Write the before and the after as two sentences with the same subject, and if the two sentences are the same, there is no story. Not a worse story: none, and no amount of shooting will produce one, because what is missing is a change of state rather than material.
On the subject version both sentences come out as a man paints numbers on seats, which is the whole diagnosis. On the version worth filming they separate: before, a man repaints seat numbers from a plan the club gave him; after, a man repaints seat numbers from memory, because he is the last copy of a record the club has lost. Same man, same brush, same morning. The second sentence is somewhere the first one is not, and everything the piece needs is now decidable, starting with the fact that the plan has to be visible on screen.
And then, against but and so
The second test works on a sequence rather than on the whole. Take the beats and read them aloud with the joins said out loud. And then is a list, but and so is a story, and the difference shows up in about four seconds of reading.
Ivo's four beats, joined the second way: he starts row 12, but the numbers are wrong, so he paints them from memory, and the club repaints the row the following week. Each beat is caused by the one before it. Joined the first way, with and then between each, the same four events sit in a row like items on a receipt, and the audience has nothing to hold because nothing is pending.
The check that follows from this is worth more than the rule. Remove a beat: if the others still work, it was not doing anything, and it is in the piece because it was expensive to get rather than because the piece needs it. Both of these tests belong to why some sequences hold and others do not, and both of them work on paper, before a camera is switched on.
What everything passes through
One change of state, and both cones hang off it
Everything shot has to reach the turn, and everything the audience leaves with comes out of it. A piece with no waist has two unrelated cones and no reason to be one film.
- Thirty years of the same August morningTrue, filmable, and not yet about anything
- The rebuilt north standWhy the plan and the seats disagree at all
- The plan from the officeHas to be on screen before the turn, or the turn is invisible
- Row 12, six in the morningThe scene the turn happens inside
- He is the record, not the planThe thing the audience now knows and did not before
- The club repaints the rowThe consequence, and the reason the piece can end
- Who is allowed to be right about a placeWhat it turns out to have been about
The waist is eleven seconds of a two minute piece. That proportion is normal and it is the argument for cutting anything that does not either reach it or come out of it.
Scene, summary, and the ratio between them
A scene has a place, a time and continuous action, and the audience is in the room for it. Summary compresses: thirty years in one line, spoken over a wide shot. Both are necessary and the ratio is what people feel as pace.
The two minutes here run 95 seconds of scene and 25 seconds of summary, and the eleven seconds in which the plan turns out to be wrong sit inside the scene rather than being described to anybody. The scene is Ivo at row 12 with the brush and the phone call from the office; the summary is the thirty years, the rebuilt north stand, and what the club is planning. Push the summary above about a third and the piece turns into an illustrated article, which is the most common shape of a documentary that nobody finishes. Pace as the thing the audience feels is the same lever pulled at a smaller scale, inside the scene rather than between scenes.
The opening is where the ratio matters most, and the rule there is a duration. Open a question in the first eight seconds, and open one the piece actually answers: the sound of a brush, and a number that does not match the seat under it. The first seconds, and what buys the next ones is a cluster rather than a line here because the cost of getting it wrong is the whole piece.
The stall is a subject that everybody agrees is interesting
The stall in this subject is specific and it is socially reinforced, which is why it lasts. A subject can be described at a table and everybody says it sounds good, because a man who has painted the same seats for thirty years genuinely is interesting. So the shoot happens. Nothing during it feels wrong, since the material is exactly what was described. The failure arrives weeks later, in an edit, as a general sense that the piece does not go anywhere, and by then it is diagnosed as an editing problem and treated with music.
What makes it a stall rather than a mistake is that the two-sentence test takes ten seconds and is never run, because the subject already passed a harder-looking test: other people liked it. A subject that survives a dinner table and fails on paper is the normal case, not the unlucky one.
The order of learning
Every stage after the first costs money to be wrong in
Structure is the only stage where being wrong is free, which is the entire reason it goes first rather than any claim about it being more important.
- freeStructure, on paperThe two sentences and the joins. Ten minutes, and free to redo
- freeThe openingEight seconds that decide whether the rest is watched
- a shootShot languageWhat the camera says. It can only say it about a story that exists
- an editRhythm in the cutThe same scene to summary lever, pulled inside a scene
- as neededThe page, or the boardPinning it down wherever somebody is about to spend money on it
The last stop is out of order on purpose: scripting and boarding are not a late stage, they are whichever stage has a budget attached to it, and on a two person documentary that can be never.
The order, and the version that costs a year
Structure first, on paper, because it is the only stage where being wrong is free. Then the opening, then what the camera is saying, then rhythm, and scriptwriting or boarding wherever the work needs the piece pinned down before money is spent on it. What the camera says without anyone speaking comes after structure rather than before it, and the reason is unglamorous: a shot can only say something about a story that already exists, and a beautiful shot of a subject with no turn in it is the most expensive thing in this pillar.
Taken backwards, the sequence has a name in every cutting room: fixing it in post. Drawing it before it is expensive is the cheap end of the same principle, and it is worth doing badly with a pen rather than well with software, because its whole value is that it is thrown away.
What to open in 2026
Fountain is the format worth knowing: screenplay formatting written as plain text, so a script is a text file that any editor opens and any version control system can read. It is free and open, and it replaced needing a proprietary application before a page could be correctly formatted. Final Draft is the paid standard that most professional production still runs on, and what it replaced was doing typewriter era formatting rules by hand.
Descript is the one that changed documentary work specifically. It transcribes the footage and lets the piece be cut by editing the transcript, so the structure can be rearranged as text before a timeline is touched, which is exactly the stage this page argues for. It replaced logging tapes and scrubbing for a line somebody half remembers, and there is a free tier large enough to test the workflow. Where it stops being enough is putting it on the page before the camera, because a transcript can only rearrange what was already said.
What none of the tests on this page can tell you is whether Ivo wants the film made, and that question is not a craft problem with a technique attached to it. The two-sentence version of his story is accurate and it is also a claim about his employer that he has to keep working for, which the piece can make and he cannot. Deciding what to do with that sits underneath everything in moving pictures, from the shoot to the export, is the same decision at the moment of pressing in light, a lens, and deciding what to leave out, and gets sharper rather than easier when the only thing recorded is somebody's voice, in recording sound, shaping it, and making music.