
Out of the box, Lightroom is too slow for culling, and three settings fix most of it. Lightroom Classic culls respectably once you stop making it render previews mid-cull and take your hand off the mouse. This is the full setup, the keyboard-only pass, and the point where even a perfectly tuned Lightroom is the wrong tool for the job.
When you arrow through fresh imports and hit that Loading… flash on every frame, Lightroom is rendering a preview on demand, decoding a RAW file while you sit there. Multiply a beat of delay by a wedding's typical 2,000 to 4,000 RAW frames and the renders alone eat serious time before you have made a single decision. That is the whole mystery of why culling takes all night in Lightroom. You have been watching a progress bar in disguise.
The stakes are bigger than annoyance. In a UK survey of 300+ wedding photographers, culling consumed 11% of total working time, with editing at 55% and shooting at 4%. Culling is a real line item in your working year, and preview strategy is the cheapest lever you will ever pull on it.
In the Import dialog, set Build Previews to Embedded & Sidecar. Your camera already rendered a full-size JPEG inside every RAW file, and this setting tells Lightroom to show you that instead of building its own. Adobe's preview documentation covers the details. Browsing becomes near-instant because there is nothing left to render. It is the same trick that made Photo Mechanic famous, which we unpack in Photo Mechanic vs Lightroom.
In Preferences, enable the option to prefer embedded previews and leave Standard-preview generation for after the cull. Otherwise Lightroom swaps its own renders in while you work, and the stutter returns.
Zooming to check focus forces a full render. If your cull is zoom-heavy, say critical-focus portrait work, build 1:1 previews overnight and cull in the morning against instant zooms. For most event culls, the embedded JPEG plus an occasional zoom-and-wait is the better trade than hours of pre-rendering.
A catalog and previews on an internal SSD beat a shoot culled directly off a spinning external. It is housekeeping, and it stacks.
Pick or reject is a binary decision. A five-star scale is an essay question asked four thousand times. Flags for the cull, stars later for album selects. The full reasoning lives in star ratings vs flags.
Go to Library, hit E for Loupe view, then turn on caps lock, which is the best Lightroom culling trick there is. With caps lock on, every flag keystroke auto-advances to the next photo. Your hands never leave the keyboard; the photos come to you.
| Key | What it does | When to use it |
|---|---|---|
| Caps Lock | Auto-advance after every flag or rating | Always, for the whole cull |
| P | Flag as pick → advance | Selected |
| X | Flag as reject → advance | Blink, blur, test frame |
| U | Unflag → advance | Undecided, keep moving |
| E / G | Loupe view / Grid view | Cull in Loupe; overview in Grid |
| N | Survey view of selected frames | Best-of-burst comparisons |
| C | Compare view, two frames side by side | Final ties |
| Z | Zoom to 100% | Focus checks only, sparingly |
Run it as two passes. Pass one, rejects only: X or nothing, no comparing, no lingering, you are removing the obvious failures from contention. Pass two, picks: work scene by scene on what survives, using N (Survey) to shrink a burst to finalists and C (Compare) to settle ties side by side. Comparison is the judgment your eye is good at. The two-pass method scales up to full wedding volume. That walkthrough is in how to cull 3,000 wedding photos fast, and the complete keymap (plus Bridge and Capture One equivalents) is in the keyboard-only culling guide.
Undecided frames get U: unflagged, auto-advanced past, dealt with later. The trap is letting later mean after the whole shoot, when you have lost the scene's context and have to re-judge from scratch. Sweep your maybes at the end of each scene instead, while the finalists are fresh, and give yourself a hard cap: if a frame survives two viewings without earning a P, it is an X. A maybe pile that grows past a handful is usually second-guessing wearing a system's clothes.
The cull is not finished until the flags become file actions. Open the Library filter bar, filter to picks. That is your edit set, straight into Develop. Then filter to rejects and deal with them in one batch, as an admin chore rather than a fresh round of decisions. Archive them out of the working catalog, or delete them if your written retention policy says so, after confirmed backups and never before. Handling rejects in bulk also sidesteps the delete-anxiety that makes photographers re-open closed decisions. The psychology of why one-by-one deleting hurts is covered in why picking photos is so hard.
Never edit during the cull. The moment you nudge one exposure slider while you are here, you have switched from a thousand two-second decisions to a handful of two-minute ones, and the session dissolves. Cull to flags, filter to picks, and only then open Develop. Culling and editing are different jobs that happen to share a window. Treat them that way, or read the complete culling guide for the workflow that keeps them apart.
Optimized Lightroom removes the waiting. It does nothing about the deciding, and deciding is the real cost. What is still missing:
None of this is a knock on Lightroom. Culling is a task it hosts rather than one it was built around. The question is volume, and at what point the leftover manual grind costs more than a tool that removes it.
So measure it. Next shoot, note the clock when you open the Library and when the last reject is filtered away. Most photographers have never timed a cull and are surprised by the answer. That one number is your decision input for everything below: multiply it by your jobs per year, price the hours at your rate, and the debate about paying for a culling tool becomes arithmetic.
The entry price of AI-assisted culling is lower than most photographers assume: Aftershoot Select runs $10/month billed annually ($120/year), and Narrative Select's Lite tier is also $10/month billed annually. Both are desktop apps that process locally and hand finished flags to Lightroom, both with unlimited images. So the switch question is whether the first-pass grind of your annual volume is worth more or less to you than roughly a hundred-and-twenty dollars a year.
Picture a photographer shooting a 20-wedding year at typical volumes, tens of thousands of frames crossing their eyes annually just to find the 400 to 800 photos each wedding delivers. For that photographer, software that automates the mechanical rejection pass costs pocket change against the hours it returns, and the only debate is which tool, which is why our culling software roundup compares them head-to-head. For a photographer shooting the occasional family session, the free fix in this article is the whole answer, and free trials (Aftershoot's 30 days, Narrative's full Ultra trial, FilterPixel's 4 free projects, all no-card) mean testing the upgrade costs nothing but curiosity when volume grows.
Our own position: Kepla is building the next step in this direction, AI culling as native apps for iPhone, iPad and Mac, where the AI proposes picks and you approve them from the couch instead of the desk, with Lightroom integration planned. It is not shipped today. There is a waitlist, free while we build, and until it ships the workflow above is the best free cull there is.
Import with Embedded & Sidecar previews so you're browsing the camera's own JPEGs instead of waiting on renders, switch to Loupe view, turn on caps lock for auto-advance, and cull with flags only: P to pick, X to reject, each keystroke advancing to the next frame. Then use the filter bar to isolate rejects and picks. The whole flow is keyboard-only. The mouse is what makes culling slow.
Flags, for the main cull. A binary pick/reject decision is faster and cognitively easier than placing every image on a 1 to 5 scale, comparative judgments beat absolute ones. Save stars for a later pass on your picks, where 4 to 5 stars can mark album and portfolio candidates.
Almost always previews. If Lightroom is rendering Standard or 1:1 previews on demand as you arrow through, every frame shows a 'Loading…' delay. Import with Embedded & Sidecar previews (or build 1:1 previews overnight before a zoom-heavy cull) and the delay largely disappears. Catalog optimization and a fast local drive help; the preview strategy is the big lever.
Not until volume makes the math obvious. If you shoot a handful of sessions a month, an optimized Lightroom flow is free and good enough. When you're regularly culling thousands of frames per job, entry AI culling tiers like Aftershoot Select or Narrative Lite (each $10/month billed annually) automate the first rejection pass, and both offer free trials (Aftershoot 30 days, Narrative a full Ultra trial), so you can test on a real shoot for nothing.
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.
It stays in your saved guides on this browser. Want it in your inbox too? One email with the link now, and the next guide when it is out. No pitches, unsubscribe with a reply.