The photo culling workflow, explained
Culling is choosing the keepers from a shoot before you edit. It sounds simple, and it is — but on a big job it's where most of your desk time goes. Here's what a good cull actually looks like, stage by stage, and where the modern tools save you hours.
Culling is choosing which photos to keep before you edit them. On a large shoot it is the longest stage of post-production. An efficient cull runs in four moves: browse fast from embedded previews, reject the obvious misses first, rate the keepers, then decide within each run of near-identical frames.
What culling is (and isn't)
Culling is selection, not editing. You are deciding which frames make the cut — the sharp ones, the ones with the right expression, the best of each moment — and cutting the rest. No sliders, no retouching; that comes later, on the frames you kept. Culling is not deleting either. In a well-built workflow, rejecting a photo marks it as out of the selection and leaves the file exactly where it was, so a decision taken at midnight can be reversed at noon. What a cull produces is not a smaller folder but a rated set: flags, stars or color labels attached to the photos, which your editor then reads as a filter. That is why the stage is worth naming at all. Everything downstream — the edit, the export, the gallery, the album — operates on the set the cull produced, and every minute you spend developing a frame that should not have survived is a minute spent twice.
Why is culling the slowest part of a shoot?
On a wedding or event you might come home with thousands of frames. Editing is creative and varied; culling is the same keep-or-cut decision, over and over. Two things drag it out — waiting for software to render each RAW, and reviewing every frame with no guidance. There is a third, quieter one: the decisions get worse as you go. Editing rewards attention, and attention is exactly what thousands of near-identical frames consume. An hour in, the frames stop looking different from each other, the threshold drifts, and you start keeping things you would have cut at the beginning — which is how a cull ends with a set nobody trusts and a second pass nobody planned. The two fixes above are fixes for that: previews that appear instantly keep you in rhythm, and a first pass that has already flagged focus and open eyes spends your attention on the frames that deserve it.
Browse fast
Get your frames on screen instantly. Dedicated cullers show the JPEG preview baked into each RAW instead of rendering the full file, so you can skim at speed. If your tool makes you wait for previews, that is the first thing to fix.
A reject pass first
Before rating anything, make one quick pass and cut only the obvious losses — closed eyes, missed focus, test frames. Clearing the dead weight first makes every later decision faster and less crowded.
Rate the keepers
Now go through what's left and mark the winners. Most tools give you three systems: flags (pick/reject), star ratings (1–5), and color labels. Pick one and be consistent — a simple pick flag is enough for most jobs; stars help when you need to rank a hero shot.
Decide within duplicates
You will have runs of near-identical frames. Put them side by side and choose the one with the sharpest eyes and best expression. A compare view turns a debate into a glance.
What can AI actually do in a photo cull?
Modern cullers add an AI first pass: they read focus, open eyes and expression on every frame so you skim with a head start and slow down only on the close calls. You still decide — the AI just points you at the likely keepers. Cullit does exactly this: sharpness, eyes and smile on every detected face. What that first pass cannot judge is worth being clear about. It sees whether a face is sharp; it does not know that the sharp frame is the one where the couple finally looked at each other. It can tell you that eyes are open in four frames of a burst; it cannot tell you which of the four is the photograph. Treat the signals as a sort order, not as a verdict: filter on them to decide what you look at first, then judge the frame yourself. The useful test of an AI cull is whether you can disagree with it in one keystroke.
What's the difference between culling and editing?
Keep them separate. Once your picks are chosen, hand them to Lightroom or Capture One as ratings written into each photo's XMP, and edit only the keepers. Culling in your editor works, but it's slower — which is why so many pros cull in a dedicated tool first.
In short
A good cull is fast browsing, a reject pass, a clear rating system, a compare view for duplicates, and — increasingly — an AI head start. Get those right and the slowest part of post stops eating your evenings.
Related
Primary sources for the standards named above:
- XMP (Extensible Metadata Platform) — Adobe, the metadata format star ratings and color labels are written into — an ISO standard (16684-1) since 2012
- Metadata basics and actions in Lightroom Classic — Adobe, states that for proprietary raw formats, XMP is stored in a separate sidecar rather than written into the file
- Exif — Exchangeable image file format for digital still cameras — CIPA, the camera metadata that records capture time, which grouping a shoot into scenes relies on