How to Find Every Shot of a Person in Your Footage

Kadir Ristjan Klade
Kadir Ristjan Klade Lead Engineer
Updated 2026-07-26

PostSlate’s face search is currently in development. The goal: identify a person once, then find every clip they appear in across your entire project, including b-roll, without scrubbing or manual logging.

Say you’re cutting a montage for a food creator named Alex, and you need every shot where Alex is on camera, across four cameras and two days of footage. You could scrub through everything and note timestamps as you go. Or you could remember that you shot “the ice cream scene” somewhere in there and search your transcripts for it, which only works if Alex actually said “ice cream” out loud.

Neither approach scales past a handful of clips. What you actually want is to search by who’s in the shot, the same way you’d search by what was said.

Why this is hard without tooling

People don’t narrate their own appearances. A subject can be on screen for the entire interview and never once say their own name where a transcript search would catch it. Finding them means either watching the footage or remembering exactly where you saw them, and memory gets worse the longer a project runs.

Most NLEs have no answer for this at all. You’re left with folder structure and hope: naming bins by person, hoping you filed every clip correctly, hoping you remember which bin has the b-roll from day two.

How face search works

PostSlate recognizes faces across your footage and groups them, so you can select a person once and pull every clip they appear in, including b-roll and cutaways where you never thought to log them. The matching happens locally, on your machine. Faces aren’t uploaded anywhere to be recognized.

Once a person is identified, that identity carries across your whole project, and across past projects too. Shot the same interview subject last year? Search for them and pull footage from both shoots at once.

Using it in your workflow

The pattern is simple: identify a person once, then search by name instead of scrubbing. Pull every clip of Alex saying anything, drop the selects into your arrangement, and keep editing. It replaces an afternoon of scrubbing with a search that takes a few seconds, and it works whether or not the person ever said their own name on camera.


Related reads:

Sign up for the waitlist →

Discuss on Hacker News

The part where you give us your email

... you get the idea, we'll get in touch once we have something