How to cull photos fast: a workflow with real timings

Culling gets fast when you stop asking every question at once. This is the workflow to hand a new second shooter: three passes, one question per pass, and a time target you can actually hold yourself to. It works on a headshot session, a two day conference and a product catalogue. Weddings have their own guide, and it is linked below.

Fast culling is an order problem, not a clicking problem

Culling is slow for a boring reason. Most photographers open the folder and ask four questions of every single frame at the same time. Is it sharp. Is it better than the one next to it. Does the client need it. Does it belong in the final gallery. Four judgments, eighteen hundred frames, and you are still on the first one when your coffee goes cold.

Speed does not come from a faster mouse or a better shortcut. It comes from asking one question at a time, in a fixed order, and refusing to answer the other three until it is their turn. That is the whole method. Everything below is just the order, and the clock.

It helps to know what this costs you. A UK survey of more than 300 wedding photographers, reported by PetaPixel, put culling at 11 percent of working time and editing at 55 percent. That was 2020, before AI tools were in anybody's workflow, so read it as shape rather than gospel. The shape is the useful part. Picking photos is not the biggest block in your week, it is the block standing in front of every other one. Nothing downstream starts until it is done, which is why a slow cull feels so much worse than the hours suggest. We put real figures on that in how long culling actually takes.

Decide how many photos you are delivering before you open the folder

If you do not set a number, you cull against an unbounded standard, and an unbounded standard has no finish line. You keep going until you get tired, which means the last hundred frames get judged by a different, crankier photographer than the first hundred.

So write the number down first, as a range, not a point. Around 90 to 110 for a family session. Around 60 for a headshot sitting where the client picks two. A commonly cited rule of thumb is 40 to 60 delivered images per hour of coverage, and for an eight to ten hour wedding that lands at the familiar 400 to 800 delivered out of 2,000 to 4,000 frames captured. Those are typical ranges rather than laws, and your own last five galleries are better evidence than anyone's average. Go and look at what you actually sent.

The number does two things. It tells you how hard to squeeze in the group pass, and it tells you when to stop. And if you have never written down what makes a frame a keeper for you specifically, do that once and reuse it forever: building your culling criteria walks through it.

The workflow, with a time target on every step

Here is the whole thing on one screen. The targets are per 1,000 frames, so scale them: a 3,000 frame event is three of these, ideally not in one sitting.

These are targets, not measurements. They are arithmetic: 25 minutes across 1,000 frames is 1.5 seconds a frame, which is one look and one key press. If a step is running long, the problem is almost always that you started answering a later step's question early.

Two notes before the detail. First, the skim is not optional and it is not a waste of five minutes. Looking at a whole shoot without deciding anything sets your standard for the day, and a standard set in advance is far more consistent than one you invent frame by frame. Second, the targets assume previews are already built. Watching a progress bar is not culling. If yours are slow, fix that before you fix anything else, and measure yourself against your own last few shoots rather than against a photographer on Instagram who claims twelve minutes.

The technical pass: one look, one key, no zooming

This pass answers one question: could this frame ever be delivered? Not is it good. Usable.

Out means the subject missed focus, the motion blur was not the plan, the flash did not fire, the eyes are shut in a frame where the eyes are the point, or the camera was photographing your shoe. Everything else stays in, including frames you are lukewarm about. Lukewarm is a pass two problem.

Three rules make the target achievable. Work full screen, one image at a time. Use one key press per frame, never a mouse. And do not zoom to 100 percent. Zooming is the single biggest time sink in this pass and it almost never changes the decision, because a frame you have to inspect at 100 percent to justify is not going in the gallery anyway. If a shot matters enough to check properly, flag it and check it in pass two, once, with the group.

If you are doing this in Lightroom and it feels like wading, the fault is usually configuration rather than the software. Smart previews, the right cache, and the loupe view instead of the grid change the experience completely. Then learn the four keys you need and nothing else, which is the short version of keyboard shortcuts for culling.

The group pass: one frame per moment

Everything that survived pass one is technically fine, which means comparing is now worth your attention. Sort so near identical frames sit next to each other, then work group by group instead of frame by frame. You did not shoot 1,000 photographs. You shot two or three hundred moments, several times each.

The rule is one frame per moment. Two only when the second frame does something genuinely different: a different expression, a different person looking at the camera, a vertical you know the client will want for print. Six versions of the same laugh in a gallery does not read as thorough. It reads as nobody chose.

When two frames are close, decide in this order: expression, then eyes, then hands, then sharpness. That order surprises people, but the technically better frame with the worse expression is the one clients quietly skip past. Give yourself five seconds. If you are still torn at five seconds, the two frames are equivalent and the choice does not matter, so take the left one and move. More on that judgment in choosing the best frame from a burst.

The story pass: look at the gaps, not the photos

The last pass is not about picking. It is about checking what is missing, and it is the pass most people skip, which is why galleries come back with a note asking whether there are any photos of the grandmother.

Put your shortlist in time order and read it as a sequence. Is there an establishing wide shot? Is there a photo of the room before anyone touched it? Did every person who was there make it in at least once? For commercial work, is every item on the brief covered, with the angles the brief actually named? This takes ten minutes and it catches the errors that cost you a reshoot.

Then stop. A shortlist is finished when it covers the story and hits your number, not when you feel confident, because you will not feel confident, and waiting for that feeling is how a 75 minute job becomes a Sunday. Everything you did not pick is still sitting on the drive exactly where it was. Nothing has to be deleted for this workflow to work, and keeping the rest is what makes the fast decisions safe: what to do with the photos you do not pick.

What changes when the shoot changes

The three passes never change. What changes is which pass eats the time, and one rule per genre that keeps it honest.

Weddings are the outlier and they deserve their own treatment, because the volume is different in kind rather than degree and the emotional stakes change how you judge a frame. That is a separate guide: how to cull wedding photos fast.

The other genre worth calling out is anything with a contract attached. On commercial and real estate work, the brief outranks your taste. Cover every required item and angle first, confirm the shortlist satisfies the deliverable, and only then argue with yourself about which of the two acceptable frames is prettier.

The four things that blow the time budget

Editing while culling. You open one frame in Develop to see whether the shadows lift, and twenty minutes vanish. Culling and editing are different jobs with different brain states. Do not mix them.

Doing pass one twice. If you find yourself back at the start second guessing frames you already handled, you are not being careful, you are being tired. The pass was fine. Go to bed.

Culling at 11pm. Research on decision fatigue suggests repeated choices degrade the quality of the ones that come later, and parts of that literature are debated, but every photographer has lived the effect. The related and better established finding is choice overload: in Iyengar and Lepper's classic 2000 jam study, more options made people less likely to choose at all and less happy with what they chose. A folder of 1,800 near identical frames is that experiment, run on you, on a weeknight.

Starting before the copy finishes. Ingest, back up, then cull. Never all three at once.

Software changes the arithmetic rather than the method, because the passes stay the same and a tool does the first one. That is the honest case for AI culling, and it is what Kepla is being built to do: the first pass on your iPhone or iPad on the drive home, then the same shoot open on your Mac for the comparisons a phone is bad at. It picks and it shortlists. It never deletes, moves or renames a file, so every original stays exactly where you put it. The picking app is not shipped yet. The booking page is live today and free while we build. And when the deadline is tonight rather than this week, there is a compressed version of all of this: culling 1,000 photos in an hour.

FAQ