You shot 4 hours. You'll use 12 minutes. Here's where the other 3:48 actually goes

Four hours of footage. Twelve minutes in the final cut. That is a 20:1 shoot ratio — 240 minutes recorded, 12 minutes used — and it sits right inside the normal range for real-world work. The other 3 hours and 48 minutes are not wasted. They are the price of finding the good parts. The question every editor lives with is where those 228 minutes actually go, because most of them are spent watching footage you already know you'll cut. This post does the arithmetic instead of quoting a marketing percentage.

What is a shoot ratio?

A shoot ratio (or shooting ratio) is how much footage you record compared with how much ends up in the finished cut, written as recorded:used. Shoot four hours and keep twelve minutes and your ratio is 20:1 — 240 minutes in, 12 minutes out. Ratios vary by genre: scripted narrative work tends to run 10:1 to 25:1, while documentary runs far higher, from 30:1 up past 80:1, because so much has to be captured to find the moments that matter (per Wikipedia's shooting-ratio entry and VashiVisuals). The higher your ratio, the more raw footage you have to watch, log, and cull before the edit even starts. That review pass — not the edit itself — is where most of the hours go. Sisyphos is a local-first desktop app that watches your raw footage, finds the best moments, and explains every pick by what's on screen — motion, sound, beat, and framing. Nothing leaves your Mac.

How much footage do editors actually use?

Less than you'd guess, and it depends on what you shoot. A tightly scripted piece might run 10:1 — ten minutes recorded for every minute used. Documentary and event work run much higher, often 30:1 to 80:1 and beyond, because the moment you need can't be staged, so you keep the camera rolling. A wedding shooter, a multi-cam sports session, a run-and-gun B-roll day — all of them push the ratio up.

So the 20:1 in the headline is not an extreme. It is a middle-of-the-road day. And it means that for every minute you keep, nineteen minutes get watched and set aside.

Where the other 3:48 actually goes

Here is the honest breakdown of a normal cull, before a single edit decision gets made:

• Watching, in real time. To find your twelve minutes, you have to see the four hours at least once. Four hours of footage takes four hours to watch. There is no fast-forward through a shot you haven't seen yet.
• Re-watching. The good takes rarely announce themselves the first time. You scrub back, compare two similar takes, and watch the near-misses twice to be sure.
• Logging. Marking in and out points, naming selects, dragging keepers into bins. This is the unpaid half of video editing, and it scales straight up with your shoot ratio.

None of that is the edit. The edit — the part you're actually good at, the part a client pays for — hasn't started yet. The 3:48 is what it costs, in hours, to reach the twelve minutes worth cutting. On a heavy shoot day it's an evening. On a wedding backlog it's a week.

How much editing time does an AI culling pass save?

This is where most articles reach for a number like "80% faster" without showing any work. So here is the actual mechanism, not a percentage.

A measured-signal pass changes what you're doing during those hours. Instead of watching four hours in real time to find the moments, you review a ranked list of candidates, each one already marked with a reason. You are reading verdicts, not scrubbing a timeline. The four hours of watch-time is the cost you stop repeating.

Say a tool watches the footage and hands you a shortlist — a ranked set of moments, each with a plain-language note like "strong motion, clean framing" for a keeper or "too similar to a better take" for a drop. Reading that list and making the call is minutes of work, not hours, because you're scanning reasons instead of watching every second unfold at 1× speed. That's a worked example, not a guarantee — your own shortlist length and how carefully you review set the real figure. But the arithmetic underneath it is simple: 240 minutes of watching you no longer have to sit through yourself.

The footage still exists. You still make every keep-or-cut decision. What changes is that you start from a reasoned shortlist instead of a blank timeline and a full memory card.

The part that matters: you still decide

A shortlist is only useful if you can trust it and argue with it. That's the difference between a culling assistant and a black box. Every pick comes with a reason. Every reject, too.

Sisyphos keeps a record of each candidate clip: whether it made the cut, its rank, and — when it was dropped — the reason in plain words. A near-duplicate falls out as "too similar to a better take." A slow stretch falls out as "too little happening." Those are causes an editor recognizes, not codes. Ask "why isn't this moment in my cut?" and you get an answer, not a shrug. You scan the rejects as fast as the keeps, because each one tells you what it was and why it lost.

Then the selects travel where you edit. Sisyphos pushes them into DaVinci Resolve as a timeline on your original files, or exports standard FCPXML and EDL for Premiere and Final Cut. What lands is a rough cut you can open and change, not a locked export.

The point of closing the gap

Cutting the watch-time out of your shoot ratio is the whole idea behind AI video culling — reviewing a ranked list instead of scrubbing hours of raw footage. If that gap is what's keeping you from your next shoot, Sisyphos is being built to close it, and it comes as a one-time license: €129 at launch, perpetual, with 12 months of updates. Bring your own AI key — cents per clip, with a hard cap. No subscription to renew, because it runs on your hardware.

If your footage has no dialogue — sport, weddings, drone, B-roll — that's exactly the case transcript tools can't touch. It's worth understanding what separates culling from editing before you pick a tool for it.

Sisyphos is in development. The workflow above — a reasoned shortlist from your raw footage, with the record of why — is what it is being built to do. Join the waitlist and you hear it first when it ships.

You stay the editor.