Resolve & Colour line · stop 09 of 14 · 32 min · members
Pulling one performer out of eight productions: face search, selects, pacing and the cut that gets a callback.
Free with an account
Membership is free: an account opens all 86 script pages. The Lab, Studio Canvas and the paid guides need the $99 pass, paid once. Already signed in on this browser? The page opens by itself.
The problem
Most of the work is finding the performer, not cutting them.
A showreel is assembled from productions the performer appeared in — often eight or ten, often full episodes, often with the performer on screen for a few minutes of each.
Done by hand, this is days of scrubbing. And it is the least valuable part of the job: locating a face in footage is mechanical, while choosing which moment shows what a casting director needs to see is not.
The approach below spends effort on the finding so that the choosing gets the time it deserves.
Stage one
Shot detection and a face pass turn hours into a candidate list.
Run scene detection across every source to get shot boundaries, then sample a frame from each shot and check whether the performer appears.
The output is a list of shots containing the performer, with timecodes. That single list is the difference between a week and an afternoon.
Expect false results in both directions and plan for them: a similar-looking actor will match, and a shot where the performer is turned away will not. The list is a starting point that removes ninety percent of the material, not a finished selection.
Stage two
You are cataloguing capability, not collecting good moments.
Watch the candidate shots and note, for each, what it demonstrates: dialogue, stillness, comedy, anger, physical work, a scene partner, a close-up.
This is the actual selection criterion. A reel needs to show range and it needs to show the performer clearly, and those are different from a shot simply being good.
Note the practical facts too — is the performer alone in frame, is the audio usable, is the shot long enough to hold. A brilliant moment in a wide two-shot where they are barely visible does not make a reel.
Stage three
The reel is watched by someone who will stop early.
Casting viewers do not watch to the end. The first ten seconds decide whether the rest happens.
So the opening is the strongest clear shot of the performer's face doing something specific — not an establishing shot, not a title card, not a montage. Face, immediately, in a moment that shows presence.
Then range, in descending order of strength. Keep it to ninety seconds or less; a two-minute reel is a ninety-second reel with the weakest material still in it.
No music underneath dialogue. It reads as decoration and it makes the performance harder to assess, which is the one thing the viewer is there to do.
Craft
A showreel is not a trailer.
The instinct is to cut fast and energetically. It works against the purpose: quick cutting shows editing, not acting.
Hold shots long enough to read a performance. Two seconds is not enough to see anything. Let a moment play, then cut.
Give the context a beat where it is needed — a line from a scene partner, a look before a reaction. A performance without its setup is not assessable, and stripping it to only the performer's frames is a common mistake.
Delivery
Every additional step loses viewers.
Name the file with the performer's name and the year. Deliver a single video with no password, no intro card longer than two seconds, and no requirement to request access.
Include a second, shorter cut if the range genuinely differs — a drama reel and a comedy reel — but do not split a single body of work into three files nobody will open.
Keep the project and the selection log. Reels get updated with each new production, and rebuilding the search from nothing every year is exactly the work this process exists to avoid.
1:1 · two hours