Film
How a reel gets cut
Forty clips in, sixty seconds out. The hard part isn't the cutting, it's deciding what never makes it to the timeline, and doing that before you fall in love with anything.
Most short-form edits fail in the first ten minutes, before a single cut is made, because everything gets dragged onto the timeline and the edit becomes an exercise in deleting things you've already grown attached to. Selection is the job. Cutting is just the paperwork afterwards.
1. Watch everything once, judging nothing
One pass, in order, at normal speed, with a notepad. Mark the timecode of anything that made you lean forward, a gesture, a light change, a moment of contact between two people. Don't rate clips. Don't trim. You're building a map of where the life is, and the only question is "did something happen here."
A second, faster pass then marks the technical facts you'll need later: which clips are shaky at the head or tail, which are out of focus, which have a subject drifting out of frame. Almost every handheld clip has an unusable first and last second, the moment you press record and the moment you reach for the button again. Note it now so you never audition those frames.
2. Select for the set, not for the clip
This is the rule that changes edits the most, and it is counter-intuitive: the best sixty seconds is not made of the sixty best seconds.
If you pick clips individually, you will pick eight variations of the same beautiful shot, because whatever you were drawn to on the day, you shot repeatedly. Played back to back they read as one shot that won't end, and the viewer's attention drops off a cliff.
So judge candidates against what's already selected, not against each other. Before a clip goes in, ask: does this differ from its neighbours in scale (wide, medium, detail), in motion (static, pan, subject-moving), and in content? If it matches its neighbour on two of those three, one of them has to go, usually the weaker one, sometimes the better one, if the weaker one is the only thing keeping the sequence varied.
Two similar shots next to each other don't add up. They cancel out, and the sequence goes flat exactly where you thought you were being generous.
3. Find the spine before you cut to music
A reel needs a shape, and there are only a few that work at this length: arrival → work → result; detail → reveal → consequence; before → during → after. Pick one, write it on a sticky note, and assign each of your selects to a position in it. Anything that doesn't belong to a position is out, no matter how good it is.
Only once the spine exists should music enter the room. Choosing the track first is seductive and it inverts the process, you end up cutting to the drop instead of to the story, and every reel you make comes out with the same rhythm regardless of subject.
4. Cut on motion, and trim to a legible frame
Two mechanical rules do most of the work of making an edit feel professional rather than assembled.
Cut during movement, not around it. If a subject's hand is travelling across frame, cut mid-travel and let the next shot continue the direction. The eye is busy tracking motion and doesn't audit the join. Cut on a static beat and the join is the most interesting thing on screen.
Land on something readable. The first frame after a cut should show a clear subject, in focus, doing something. It is remarkably easy to trim to a frame that is technically the start of the action but visually is an empty street, or the back of someone's head, or the half-second where a person is being directed and hasn't started performing yet. If the first frame doesn't communicate, push the cut later until it does.
Keep the action off the edges too. On vertical delivery, the top and bottom of the frame get covered by interface, the caption bar, the profile row, the button stack. Anything meaningful within roughly the outer 12% of the frame will be sat on by something. Reframe or scale in slightly rather than hope.
5. Reframe deliberately, once
Cutting horizontal footage for vertical delivery is not a crop, it's a second composition. Decide per shot where the subject should sit, usually a third from the top for a standing figure, centre for a detail, and set it as a static reframe. Constant slow drift on every shot ("Ken Burns everything") is the fastest way to make a reel look automated. Move the frame when the movement means something: to follow a subject, or to reveal.
6. Grade last, and measure rather than trust
Leave the look until the cut locks, for a practical reason: a grade tuned to a sequence you then reorder is a grade you will do twice.
When you do get there, the one habit worth building is checking the actual decoded picture instead of trusting the settings. Modern cameras produce log footage in wide colour spaces, and it is entirely possible to have a technically correct conversion applied, right transform, right table, everything by the book, and still be looking at a picture that's wrong, because the footage was tagged as one thing and interpreted as another somewhere upstream. The settings panel will tell you it's fine. The picture will tell you it isn't.
So: put a scope on it. Check that black sits where black should sit and that skin lands in a plausible range. Compare against a reference frame you trust. If those two disagree with what the interface claims, believe the measurement, that discrepancy is exactly where the hours disappear on a bad day.
7. Captions and the last pass
Most viewers watch without sound, so captions are structural, not decoration. Keep them to a few words per card, place them clear of the interface zones, and let them punctuate the cut rather than narrate over the top of it.
Then watch the whole thing once at full size, once on a phone, and once with the sound off. Each pass catches a different category of mistake: pacing on the big screen, framing on the phone, and, with the sound off, every place where the music was doing the work you thought the pictures were doing.