Culling is the quiet bottleneck of every shoot. You come home from a wedding with 2,500 frames, or back from a weekend hike with 800, and the editing cannot start until you decide which photos are worth keeping. Most photographers dread this step more than the edit itself, because it feels like janitorial work: scrolling, squinting, guessing whether frame 41 is a hair sharper than frame 42. This guide lays out a fast, repeatable way to cull, explains what an automated scorer can actually judge for you, and shows how to do the whole pass on your own computer without ever risking a keeper.
What culling really means (and what it is not)
Culling is the act of choosing which photos survive to the editing stage. It is a filtering decision, not a destructive one. A clean mental model separates three jobs that beginners often blur together:
- Culling picks the frames worth your attention. Nothing is changed or removed, you are only sorting keepers from rejects.
- Deleting permanently removes files. This should be a rare, deliberate action, done only after you are certain, never as part of a first pass.
- Editing develops the survivors. Exposure, color, and crop belong here, after the field has been narrowed.
Keep those three separate and culling stops feeling scary. A reject flag is reversible. You can revisit it tomorrow, change your mind, and lose nothing. The moment you treat culling as a delete-as-you-go chore, every decision carries weight it should not, and you slow to a crawl.
The two-pass method that still beats everything
Speed in culling comes from making cheap decisions first and expensive decisions last. A single pass where you agonize over every frame is the slowest possible approach. Split the work instead:
- First pass, reject the obvious. Move quickly and only look for disqualifiers: badly missed focus, eyes fully closed, a blown-out sky with no detail, an accidental shutter of the floor. Do not compare similar frames yet. If a photo is clearly unusable, flag it as a reject and move on. Aim for one or two seconds per image.
- Second pass, compare the survivors. Now group the near-identical frames, the bursts and the reshoots of the same moment, and pick the single best from each cluster. This is where you zoom to 100 percent, check the eyes, and choose the sharpest expression. It is slower, but you are only doing it on the handful of frames that earned it.
Two focused passes almost always finish faster than one anxious pass, because the first pass removes the noise before you spend real attention. On a large event shoot, that split can turn an evening of culling into an hour.
Where the time actually goes
If you watch where minutes disappear during a cull, four culprits dominate. Naming them helps, because each one has a shortcut.
| Time sink | Why it is slow | Faster approach |
|---|---|---|
| Bursts and motor-drive sequences | Ten frames of the same second, all nearly identical | Judge the group as one unit, keep the single best frame |
| Soft focus you cannot see at fit-to-screen | A shrunk preview hides missed focus | Check sharpness at 100 percent only on shortlisted frames |
| Closed or half-closed eyes in group shots | Easy to miss on a small thumbnail | Compare the two or three best frames side by side for open eyes |
| Near-duplicate reshoots | You photographed the same setup twice | Cluster by similarity, then choose once |
Notice that three of those four are comparison problems, not quality problems. The slow part is rarely deciding whether one photo is good. It is deciding which of five very similar photos is the best. Any tool that clusters look-alikes and surfaces the strongest candidate attacks the real bottleneck.
How AI scoring speeds up the first pass
Automated culling has a narrow, honest job: handle the mechanical first pass so your eyes are fresh for the judgment calls. A local culling engine looks at each frame and scores a few objective things that a computer can measure reliably:
- Sharpness, estimated from how much fine detail and edge contrast a frame carries, so genuinely soft shots float to the top of the reject pile.
- Exposure, by counting how many pixels are blown to pure white or crushed to pure black, which flags frames with no recoverable highlight or shadow detail.
- Closed eyes, using a small eye-state model, so the obvious blinks in a portrait get flagged before you ever open the folder.
- Best frame in a burst, by grouping shots taken within a short time window that also look alike, then ranking them so the strongest of the sequence is easy to keep.
Be clear about the limits. A score cannot know that the blurry frame is the one shot you got of the bride laughing, or that a technically perfect photo is boring. Focus, exposure, and open eyes are measurable. Emotion, timing, and story are not. Treat the scores as a triage nurse, not a judge. They tell you where to look first, and you make every real decision. Good culling software proposes, it does not decide, and it never deletes anything on your behalf.
A faster culling workflow, step by step
- Import and let the library index. Bring the shoot into your photo organizing software and let it read the files first. A tool that decodes your camera's RAW files directly, rather than making you convert them, saves a whole step here.
- Run an automated first pass. Let the culling engine flag soft focus, closed eyes, and blown exposure, and let it group your bursts. Review its suggestions rather than trusting them blindly.
- Sweep the rejects quickly. Scan the flagged pile to rescue any false positives, the intentional silhouette read as underexposed, the artful motion blur read as soft. Confirm the rest.
- Compare within each burst. For every group of look-alikes, open the top two or three candidates, zoom to 100 percent, and keep one.
- Star the keepers. Mark the frames that graduate to editing. Everything else stays on disk, flagged, costing you nothing and available if you change your mind.
- Only then consider deleting. If you truly want to reclaim space, review the reject pile one last time, and remove files as a separate, deliberate act, not as a reflex during the cull.
Common culling mistakes to avoid
- Deleting as you go. It makes every decision permanent and turns a fast filter into a slow, stressful commitment. Flag first, delete later, if ever.
- Judging focus at fit-to-screen. A downscaled preview can make a soft frame look sharp. Check critical focus at 100 percent, but only on the frames that already survived the first pass.
- Culling tired. Fatigue makes you either ruthless or careless. If the numbers are huge, split the session, and let the automated pass carry the mechanical load.
- Trusting the score over your eyes. The sharpest frame is not always the best frame. Use the ranking to save time, then let the moment win when it matters.
- Skipping the burst comparison. Keeping every frame of a ten-shot burst is not culling, it is postponing the decision. Choose one.
Culling locally versus culling in the cloud
A growing number of culling tools run your images through their own servers, or require an account and an upload before they will score anything. For a private commission, that raises real questions. A wedding or a client portrait session is sensitive material, and sending thousands of unedited frames to a third party before you have even chosen the keepers is a bigger decision than most photographers pause to make.
Culling on your own machine sidesteps the problem entirely. When the scoring runs on your CPU or GPU, nothing is uploaded, nothing waits on a connection, and the client's photos never leave your computer. You also get to work on a plane, in a hotel with bad wifi, or anywhere the internet is not cooperating. Local culling is both a privacy choice and a reliability one.
How Memora handles the cull
Memora is a Windows photo manager that runs every step described above on your own computer. Its AI Culling feature (part of Memora Pro) scores each frame for subject sharpness, eye focus, whether eyes are open, and exposure, then groups bursts and surfaces the strongest frame in each sequence. Four sliders let you set how strict each judgment is, and the thresholds adapt to the folder you are working in, so a low-light reception is not judged by the same standard as a bright outdoor ceremony. Crucially, Memora never deletes a photo for you. It flags rejects and proposes the best frames, and every final call stays yours.
The pieces around the cull are local too. Memora decodes RAW files from more than a thousand cameras directly, so there is no conversion step before you can review a shoot, and its RAW support covers formats like CR3, ARW, NEF, and DNG. A built-in duplicate finder groups exact and near-identical copies for cleanup after the cull. And once the keepers are chosen and edited, AI semantic search lets you find any of them later by describing what is in the frame, all without a single photo ever being sent to the cloud.
Culling will always involve your judgment, and that is exactly as it should be. The goal is not to hand the decision to a machine, but to clear away the mechanical drudgery so the decision that is left is the interesting one. Handle the obvious rejects fast, compare the survivors with care, keep everything reversible until you are sure, and the slowest part of your workflow stops being the part you avoid.
Ready to try a faster cull on your next shoot? Learn more about Memora for Windows and how it keeps your whole library, culling included, private and on your own machine.