The Three-Pass Rough Cut Method We Use

An assembly arrives looking like a film and behaving like nothing of the kind. Every selected take sits in script order, the runtime runs an hour or two past where the picture will eventually land, and the sound is whatever the production mixer happened to catch on the day. It reads as a movie and is wrong in almost every particular — which is exactly what it should be at this stage.

The trouble starts when the next stage, the rough cut, gets treated as a single job. Asked to settle structure, performance, and sound all at once, an editor ends up re-cutting the same scene three times for three unrelated reasons. So we don't work that way. We run the rough cut as three deliberate passes, each with one question to answer, and we refuse to let the later questions contaminate the earlier ones.

One pass asks for too many decisions at once

The offline edit has a familiar shape: assembly, rough cut, final cut. The rough cut is the second of those three stages, and the first point at which the film starts to resemble its finished self. But "rough cut" is a generous label. As IndieWire's postproduction glossary notes, the term is a catch-all for anything that isn't a final cut, not a precise step with its own rules.

That vagueness is the thing we are working against. When one pass is expected to fix scene order and trim a performance and smooth a transition all together, the decisions interfere with each other. You tighten a reaction to two frames, then cut the whole scene in the next session, and the tightening is gone with it. The work was real and it evaporated. Every change we make should survive until picture lock; the only reliable way we have found to protect that is to fix one kind of thing at a time.

First pass — does the order hold?

The first pass works at the altitude of scenes and sequences, never frames. We start from the assembly and ask a single question: does the story stand up when nothing is polished? Scenes move. Scenes get cut. A sequence that read cleanly on the page collapses on screen, and we find that out now, while moving it costs nothing.

We deliberately ignore performance nuance here. A flat reading or a slack pause is not a first-pass concern, because the scene it lives in might not survive the week. The first cut is famously long, often as much as two hours past the final running time, and that length is useful — it means the structural questions are still open, not yet foreclosed by trimming. We leave the picture baggy on purpose until the spine is right.

Second pass — does each scene breathe?

Only once the order holds do we go inside the scenes. The second pass is about performance and rhythm: which take, which reaction, how long to hold before the cut, where the real out-point sits. This is the craft people picture when they imagine editing — the frame-level decisions that make a moment land or fall flat.

It's also work we'd have thrown away had we done it first. Because the structure is settled, every trim we make here is permanent; we're no longer hedging against a scene disappearing underneath us. We note continuity snags and room tone gaps as we pass through, but we don't treat them yet. The question stays narrow on purpose: does this scene, in this position, breathe the way it should?

Third pass — do the seams disappear?

The last pass is led by sound. Picture and sound do not have to break on the same frame, and most of the time they shouldn't. We go through with L-cuts and audio bridges — letting dialogue or ambience from one scene carry into the next, using a sound to pull a viewer across a cut they would otherwise feel.

These moves only work once structure and performance are fixed, because an L-cut laid across a scene boundary breaks the instant that boundary moves. Saving it for last isn't a matter of taste. It's a matter of dependency. When the seams stop announcing themselves, the rough cut is finally ready to show the director — coherent enough to react to, loose enough to keep changing.

A short note on order of operations

The method's real value is defensive. It stops us re-deciding the same scene three times under three different pressures, and it keeps each decision alive until picture lock. None of this is exotic — just an order of operations applied with a little discipline. It matters most on the sound pass, where the work is quietest and the easiest to undo by accident — which is why we save the studio's sound attention for last, when there is finally something stable to listen against. More of how we think about the cut lives across the journal, including our documentary work.