Culling time benchmarks: a number to beat
Most photographers cannot tell you how long their last cull took, only that it took too long. This article is the other half of that problem: a target rate to aim at, and a simple way to find out what your real rate is. Three shoots, one stopwatch, and a unit that lets you compare a wedding to a Tuesday headshot session.
Hours per shoot is the wrong unit
Ask around and you get answers like "about three hours" or "most of a Sunday." Those are not measurements, they are moods. They also cannot be compared, because the two photographers shot different numbers of frames, and one of them counted the hour spent waiting for previews while the other did not.
The fix is boring and it works. Stop counting hours per shoot and start counting seconds per frame. One number, one unit, comparable across a wedding, a newborn session and a two hour corporate job. If you prefer bigger numbers, use minutes per 1,000 frames, which is the same measurement with the decimal point moved and is easier to say out loud. Fifty minutes per 1,000 frames is three seconds a frame.
Why the unit matters: frame count is the thing you control at the camera, and seconds per frame is the thing you control at the desk. Roll them into one figure and you cannot tell which one moved. A photographer whose wedding cull went from four hours to three might have gotten faster, or might simply have shot 800 fewer frames. Same headline, completely different lesson.
If you want the arithmetic underneath all of this, the formula and the overhead nobody counts, we laid it out in how long culling actually takes. This piece assumes you accept the formula and want a target.
Starting lines, by genre
Read the caveat before the table. These are not research figures. Nobody has published a study of culling speed by genre, and any site that gives you one to two decimal places is making it up. What follows is a set of reasonable opening targets, picked so that a photographer with a working system can hit them and a photographer who zooms to 100 percent on every frame cannot. Treat them like a suggested running pace: useful on day one, replaced by your own data by week three.
| Shoot type | Starting target | Same number, per frame | Why it sits there |
|---|---|---|---|
| Wedding, full day | 50 min per 1,000 frames | about 3 sec | Long sequences and repeated setups, so a lot of frames are decided in a glance |
| Portrait or family session | 60 min per 1,000 | about 3.5 sec | Fewer frames, but nearly every one has a face and a fair chance of making the gallery |
| Corporate or event coverage | 40 min per 1,000 | about 2.5 sec | You need a spread of moments, not the single best version of each one |
| Sport or concert | 35 min per 1,000 | about 2 sec | Burst heavy. Focus and peak action decide most frames without much deliberation |
| Newborn or product | 75 min per 1,000 | about 4.5 sec | Small sets, high stakes per frame, real comparison work between near identical setups |
| Real estate | 90 min per 1,000 | about 5.5 sec | The decision is which bracketed sequence to blend, which is slower per frame by nature |
Suggested starting points only, not measured or published data. Use them until your own three shoot average replaces them.
Put a volume on it. Industry norms put a full eight to ten hour wedding day at 2,000 to 4,000 RAW frames, with 400 to 800 images delivered. At the wedding starting line that is roughly one hour forty to three hours twenty of decision time. If your wedding cull consistently runs past four hours, the cause is far more likely to be your process than your genre.
The closest thing to hard data in this area is old and indirect. A UK survey of 300 plus wedding photographers, reported by PetaPixel in 2020, found culling took 11 percent of working time against 55 percent for editing. That is a share of a whole year, from before the current tools existed, and it says nothing about your Sunday. It is context, not a benchmark. We pulled the full split apart in where photographers' time actually goes.
The three shoot measurement, exactly how to run it
One timed cull tells you almost nothing. Shoots differ, your energy differs, and the shoot you happen to time is usually the one you concentrate hardest on. Three shoots in a row is the smallest number that shows a pattern instead of a mood.
- Define what counts as culling. The clock starts when the first frame is on screen and you are making decisions. It stops when you have a shortlist you would hand to editing. Import, backup and preview building are real time, but they are not decision time, so they get their own line.
- Time the machine separately. Keep a second note for waiting: previews rendering, spinners, the app catching up. An honest estimate is enough. This is the number a faster computer or a smarter preview strategy fixes, and it is worth knowing before you buy either one.
- Change nothing during the three shoots. Same software, same shortcuts, same room, same time of day if you can manage it. You are measuring the workflow you have, not the one you want. Improvements come after the baseline, never during it.
- Cull in one sitting where the shoot allows. Split sessions inflate the total, because you reload the whole shoot into your head every time you come back. If you must split it, write down how many sittings it took.
- Do the division at the end, not after shoot one. Log the raw figures each time and only work out your rate when all three are in. Otherwise you spend shoot two and shoot three chasing shoot one, and you have measured your competitive streak rather than your workflow.
A phone timer is fine. What matters is that the same rule starts it and stops it every time.
Five columns, on the back of an index card
Any log longer than this and you will stop filling it in by the second shoot.
| Column | What goes in it | Example |
|---|---|---|
| Shoot | Type and date | Wedding, Aug 8 |
| Frames in | Card count after import | 3,120 |
| Clock time | Decision time only, in minutes | 142 |
| Picks out | Frames on the shortlist | 690 |
| Waiting | Rough machine time, in minutes | 25 |
Two numbers fall out of that. Seconds per frame is clock time in minutes, times 60, divided by frames in. In the example that is 142 times 60, divided by 3,120, which is about 2.7 seconds a frame: comfortably inside the wedding starting line. Pick rate is picks out divided by frames in, about 22 percent here.
Track both, or you will optimize the wrong one. A cull that hits its time target while passing 40 percent of the card forward has not saved you an evening, it has moved the evening downstream into editing, where every surviving frame costs several times more per minute. The handoff between those two jobs is the whole subject of how culling and editing fit together, and it is where fast culls quietly turn into slow months.
Nothing on the shortlist is a verdict on the rest of the card. The frames you did not pick stay exactly where they are, and a shortlist you can walk back is what makes a fast pass psychologically possible in the first place.
How to read three numbers without fooling yourself
Use the middle one, not the best one
The fastest of your three is the number you will want to quote to other photographers. Use the middle one instead. It is the rate you can repeat on an ordinary Tuesday in November, and repeatable is the only kind of number worth planning a season around.
Treat small gaps as noise
Shoots vary for reasons that have nothing to do with your workflow: how well you shot, how tired you are, how much of the day sat in one lighting scene. As a rough rule, if your three figures land within a fifth of each other, treat them as one number. Chasing a difference smaller than that is measuring the weather.
Keep your time and the machine's time apart
If waiting is a third of your total, you do not have a decision speed problem and no amount of discipline will fix it. You have a preview problem, and it is solved by hardware or by building previews before you sit down, not by trying harder.
Re-measure after one change, not five
Flags instead of stars is a change. New software is a change. A second screen is a change. Make all three at once and you will know your total moved without knowing why, which leaves you paying for two things that did nothing. One change, three shoots, compare. Swapping star ratings for a straight in or out flag is the cheapest one to test first, and usually the biggest single move on the list.
You missed the target. Which kind of slow are you?
A rate that is worse than the starting line is not a character flaw, it is a symptom, and there are only really three shapes it comes in.
Fast on most frames, stuck on a few
This is the near duplicate stall. The seconds do not accumulate evenly, they pile onto the six frames of the same moment where nothing is obviously wrong with any of them. That is a criteria problem, not a speed problem. Decide in advance what breaks a tie for your work, expression over sharpness or whichever way round it goes, and the stall mostly stops. Writing your criteria down takes twenty minutes and pays for itself in one wedding.
Evenly slow on everything
Usually the interface, not the eye. A five point star scale asks a five way question about every frame when a two way question would do, and any shortcut you still have to think about costs more than the keystroke saves. Fix the input method before you blame your judgment.
Good rate, ruined evening
If your log shows a healthy seconds per frame and your life still feels eaten, the cull was never the problem. When it happens is. Decisions made on a card dump at midnight after a ten hour shooting day are slower and more anxious than the identical decisions made on Sunday morning, and the delay between the shoot and the shortlist has costs of its own, which we added up in the hidden cost of a slow cull.
What you actually do with the number
A benchmark you never use is a hobby. Yours earns its keep in three places.
Capacity. Multiply your rate by the frames a season will bring and you get an honest answer to whether you can take one more booking. At three seconds a frame, a 3,000 frame wedding is two and a half hours of decisions before a single edit. Twenty five of those is most of a working month, and it is the difference between a full calendar and a full calendar you resent.
Pricing and turnaround. Once you know your rate you can quote a delivery date from arithmetic instead of optimism, and you can see what your post production time is really worth per hour. That hourly figure is the foundation under pricing for your editing time, and most photographers find it lower than they assumed.
Judging software honestly. An AI first pass changes the unit itself. You stop judging every frame from zero and start approving a proposed shortlist, so the right question is not "did the demo look fast" but "what did my seconds per frame do across three real shoots, and did my pick rate stay where I want it." Run the same three shoot measurement during a free trial and you have an answer instead of an impression. We went through the realistic size of that gain in how much time AI culling actually saves.
Where our own number will have to prove itself
Kepla clears the obvious misses and hands you the clean frames. Nothing is ever taken away: every frame it did not pick stays where it is, one tap from the ones it did. The picking app for iPhone, iPad and Mac is still being built, so we are not asking anyone to take a speed claim on faith. When it lands, measure it against the baseline you just built. In the meantime the booking page is live and free while we build.
FAQ
How long should it take to cull a wedding?
As a starting point, aim for about 50 minutes per 1,000 frames, which is roughly three seconds a frame. Industry norms put a full wedding day at 2,000 to 4,000 RAW frames, so that target works out at roughly one hour forty to three hours twenty of decision time, not counting import, preview building or breaks. Measure three weddings before you decide whether that target is right for you.
What is a good seconds per frame for culling?
It depends on the genre, and there is no published study to point at. Two to three seconds a frame is a sensible opening target for coverage work like weddings and events, four to five for genres where frames are few and each one matters, such as newborn or product work. The useful number is not the benchmark, it is your own average across three consecutive shoots.
How do I measure my own culling speed?
Start a timer when the first frame is on screen and stop it when you have a shortlist. Note frames in, minutes on the clock, and how many you picked. Do that for three shoots in a row without changing anything about your setup, then divide total minutes by total frames. Keep waiting time for previews on a separate line so you can tell your speed from your computer's.
Is faster culling always better?
No. Speed only counts if the shortlist is still one you would defend. Watch your pick rate alongside your time: a fast pass that sends 40 percent of the card into editing has moved the work rather than saved it, because editing costs far more per frame than deciding does. A good cull is fast and tight at the same time, and tight is the harder half.
Does AI culling change the benchmark?
It changes what you are timing. Instead of judging every frame from zero you are reviewing a proposed shortlist, so the honest test is your seconds per frame across three real shoots with the tool, compared with the same three shoots done your usual way. Most free trials run long enough to do exactly that, and the disagreements between you and the software tell you more than the total.
Should I count preview rendering as culling time?
Keep it on a separate line. It is real time out of your evening, but it is fixed by hardware, storage or building previews before you sit down, not by better decisions. Mixing it into your rate hides both problems. Many photographers who time the two separately for the first time find that the machine, not their judgment, owns a surprising share of the session.
Never sort photos at 1 AM again.
Kepla for Mac clears the obvious misses from a card, names the reason on every frame it sets aside, and leaves the choosing to you. Nothing is ever deleted, moved or renamed. Free through the private preview · the first hundred photographers keep it at $99 a year.