Culling time by shoot size: why the curve bends
Ask a photographer what a bigger job will cost them and most of us scale it in our heads: twice the frames, twice the evening. It almost never lands there. Small shoots cost more per frame than they should, large ones cost more than the arithmetic predicts, and the reason is different at each end. Here is the shape of that curve, and how to flatten it.
The arithmetic is right, and it still under-predicts
Culling time starts as a simple sum: frames, times the seconds each decision takes, plus the overhead nobody counts. We pulled that formula apart in how long culling actually takes, and it holds up. It is also a straight line. Double the frames and the line says double the time.
The line is accurate inside a narrow band, roughly the size of shoot you handle every week. Step outside that band in either direction and it starts to lie, because the two inputs are not independent of shoot size. At the small end, the overhead term stops being a rounding error and becomes most of the job. At the large end, the seconds-per-decision term climbs while you work, so the rate you measured on a portrait session is not the rate you will get on a wedding.
That is the whole point of this article. If you want a rate to aim at and a stopwatch method for finding your own, that is a separate job and it lives in our culling time benchmarks. What follows is what happens to your rate, whatever it turns out to be, when the frame count moves a long way in either direction.
The small shoot pays a fixed toll
Before a single decision gets made, a shoot costs you a fixed block of work. Cards offloaded and verified. Import and a backup you trust. Previews built. The client's brief and the shot list back in your head. A naming convention applied. None of that scales with frame count. It is close to the same block of minutes on a 200-frame mini session as on a 3,000-frame wedding day.
Put your own figure on it, because this is one of the few numbers here you can actually measure. Time the block once, from card in the reader to the first frame on screen with you ready to judge it. Whatever it comes to, it is a flat toll. On a wedding it disappears into the evening. On a 200-frame session it can outweigh every decision you make, which is why per-frame rates on small jobs look terrible and why that is not a discipline problem.
The consequence photographers miss is the mini-session day. Eight galleries of 150 frames is not a 1,200-frame cull. It is eight separate culls, each with its own toll, its own client, and its own context to load. Batch them into one sitting and you pay the setup once instead of eight times across a week. The same logic runs through high-volume graduation and prom days, where the count is enormous but each individual set is small.
Where the curve bends: comparison, not frames
Now the other end. A frame with no near neighbor is cheap: it is a yes or a no and you move on. Frames do not arrive that way on a big shoot. They arrive in clusters, the burst on the aisle, the four takes of the family formal, the fourteen versions of a laugh at the speeches. A cluster is not judged frame by frame. It is judged against itself.
That is the tax that bends the curve. The cost of a cluster depends on how many frames are in it, not on how many frames are in the shoot, and big shoots have both more clusters and bigger ones. Three clusters of four is a quick job. One cluster of twelve is not, because the temptation is to compare everything to everything, and that work grows far faster than the frame count does.
There is a method fix for this and it is the single highest-leverage habit in a long cull. Hold a champion. Move through the cluster once, keep one current best in your head, and compare each new frame only against that champion, never against the whole set. One pass, one comparison per frame, and the cluster costs what its size says it should rather than its size squared. The judgment side of that, which frame actually deserves to be the champion, is in choosing the best frame from a burst.
Your first-pass method matters here too. Marking failures leaves every technically acceptable near-duplicate standing, which is exactly the pile that grows with shoot size. Marking keepers does not. That trade is worked through in cull in versus cull out.
The second bend is you
Hour three is not hour one. Research on decision fatigue suggests that a long run of choices degrades the quality of the choices at the end of it, and while parts of that literature are debated, every photographer who has culled a wedding past midnight recognizes the effect. Your seconds per decision is not a constant. It is a number that drifts while you sit there.
It rarely drifts in an obvious way. Standards slide, so the reception gets judged by a looser rule than the ceremony did. Second-guessing creeps in, and a shoot you culled once gets culled one and a half times. The sitting breaks into three sittings across three evenings, and each restart costs you the reload: which chapter you were in, which family group is still missing, what you had already decided about the light in the room.
Scale makes this worse in a way frame count alone does not capture. A full wedding day is thousands of individual calls before a single edit, a load we counted out in three thousand decisions per wedding, and the quality of call number 2,800 is not the quality of call number 12. The practical read: past a certain size, the constraint stops being your software and becomes your attention.
Two shooters is more than two piles
Adding a second photographer looks like the cleanest possible scaling test. Same day, roughly double the frames. It is not, and the reason is the comparison tax again.
Both cameras covered the ceremony. Both covered the first dance, from different sides, at overlapping moments. So the near-duplicates are no longer inside one burst in one folder, they are spread across two folders shot by two people with two ideas about the moment. To judge them you have to merge by capture time first, and only then can you compare, which means a chunk of the work has to happen before any decision is possible.
Meanwhile the output barely moves. Commonly cited industry ranges put a full wedding day at 2,000 to 4,000 RAW frames with 400 to 800 images delivered, and a second shooter pushes the top of that capture range without changing what the couple receives. More in, the same out, so the ratio of set aside to picked gets steeper and every extra frame is one more thing to compare rather than one more thing to deliver. The workflow for keeping that sane, including whose frame wins when both are good, is in culling second shooter photos.
What actually changes at each size
Here is the curve as a table. It deliberately does not give you times, because a time only means something next to your own measured rate. It gives you the thing that dominates the clock at each size, and the thing that gives way first.
| Frames on the card | What eats the clock | What breaks first |
|---|---|---|
| A few hundred, mini session or headshot round | Setup, import, previews, loading the client's brief | Overhead. Decision time is the smaller half of the evening |
| Under a thousand, portrait or family session | Straight decision time, a handful of tight comparisons | Not much. This is the band the straight line describes well |
| One to two thousand, event or engagement | Duplicate clusters, plus a coverage check at the end | The single sitting. The cull starts needing stages |
| Two to four thousand, a full wedding day | Comparison inside clusters, chapter coverage, re-reviews | Attention. Standards drift in the last hour |
| Five thousand and up, two shooters or a school day | Cross-folder duplicates, merging, sheer scale | The method. Anything you were doing by feel stops working |
Read down the last column and the lesson is that each size fails differently, so each size needs a different fix. Buying a faster machine solves the top row and does nothing for the bottom one. Tightening your criteria solves the bottom row and does nothing for the top. Work out which row you are actually in before you spend money on the wrong one.
How to flatten the curve
Fix the batch, not the shoot. Stop culling weddings and start culling chapters. Getting ready, ceremony, portraits, reception, each treated as its own small job with its own start and stop. A 400-frame chapter behaves the same whether it came out of a 2,000-frame day or a 6,000-frame one, which drags a huge shoot back into the band where your rate is honest.
Give overhead its own line. Card offload, previews and import go in a separate column from decision time. Mixed together they hide each other, and you end up buying a faster laptop to fix a habit, or fixing a habit to solve a slow drive.
Decide the delivery target before the first frame is on screen. An open-ended cull expands to fill the night. A cull with a number attached to it ends. Industry norms put a wedding delivery at 400 to 800 images, so pick your figure inside that range and let it decide when the chapter is finished.
Keep comparisons linear. One champion per cluster, one pass, no going back for a third look at a frame you already lost. This is the habit that stops the top end of the curve from running away.
Spend your best attention on the hardest chapter. Do the ceremony while you are sharp. Leave the mechanical passes, the coverage check, the naming, for the hour when you are not.
Shrink the input. Everything above manages the pile. The other lever is arriving at a smaller one, which is partly shooting discipline and partly handing the obvious calls to software so that the comparison stage starts from a shortlist instead of a card dump. That is the job Kepla is being built for: a first pass on the iPhone or iPad on the drive home, the same shoot picked up later on the Mac for the frames that need a big screen, and nothing ever deleted, moved or renamed, so every frame it set aside is still one tap away. The picking app is still in development. The booking page is live and free today.
FAQ
Does culling time double when the frame count doubles?
Usually it more than doubles. The decision count doubles, but the comparison work inside bursts grows faster than that, and your accuracy per decision falls as the session runs long. A shoot twice the size of your usual job is better planned as more than twice the time, split across separate sittings, than as one very long evening that quietly turns into two.
Why does a small session feel slow for so few photos?
Because of fixed costs. Offloading cards, building previews, backing up and reloading the client's brief cost roughly the same on a 200-frame session as on a 3,000-frame wedding. On the small job that block of work is most of the evening, so your per-frame rate looks bad through no fault of your judgment. Batch small sessions into one sitting and you pay the setup once.
How long should a 5,000-photo wedding take to cull?
There is no honest single number, and expect the per-frame rate to be worse than on a 2,000-frame job rather than the same. Measure your rate on a shoot you cull often, then plan the big one in chapters at that rate and add margin for merging two shooters. Our benchmarks article covers how to run that measurement properly.
Is it faster to cull a big shoot in one sitting or several?
Several, once the shoot is large enough that your judgment starts drifting. The catch is that every restart costs you a reload of where you were and what you had already decided. Split by chapter rather than by clock, so each sitting has a natural start and finish, and write down what is still missing before you stop.
Does a second shooter double the culling work?
It adds more than the frame count suggests. The same moments are now covered twice from two angles, so the near-duplicates sit across two folders and have to be merged by capture time before you can compare them at all. The delivered gallery stays about the same size, which means more of the extra frames become comparison work rather than deliverables.
Does AI culling change the shape of the curve?
It mostly attacks the input. Handing the obvious technical calls and duplicate grouping to software means the comparison stage starts from a shortlist instead of everything on the card, which is where the curve bends hardest. Fixed overhead and your own taste decisions do not go away, so treat it as a smaller pile to judge rather than a finished gallery.
The big shoots are the ones that eat the week
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.