Lightroom Assisted Culling review: an honest look

Adobe now picks photos for you inside Lightroom. This is a review of what Assisted Culling actually measures, what it puts in the reject pile, and which photographers it genuinely suits. No lab test and no accuracy percentage: Adobe's documented behavior read closely, plus a free method you can run tonight on a shoot you already culled by hand.

The short version

Assisted Culling analyzes a folder or an album, scores every frame against three technical criteria, and hands back a proposed set of Selects and a proposed set of Rejects. Adobe labels it Early Access, and it exists in both of Adobe's photo applications.

That last point matters more than it sounds. Lightroom Classic and Lightroom are two different products, not two names for one app. Classic is the desktop catalog most wedding and event photographers run. Lightroom is the cloud based app that also runs on a phone and tablet. Assisted Culling appears in both, in different places, with slightly different labels on the same controls. Everything below comes from Adobe's own Assisted Culling documentation, checked in August 2026, and it is worth reading before you form an opinion from a forum thread.

What it is not: a judge of expression, timing or story. Adobe does not claim it is, and no honest review should imply otherwise. It is a fast, tireless technical screen, and a technical screen is a real thing to want at midnight.

Switching it on in each app

In Lightroom Classic you turn it on first. Adobe's instruction is to select Catalog Settings > Metadata > Assisted Culling, or to use the drop-down beside the Identity Plate and turn on Auto Analysis. After that, the Assisted Culling panel appears in the Library module once the analysis has finished on the folder or collection you pointed it at.

Classic can also run it during import. Adobe documents opening the Import window, choosing a folder from the Source menu, and selecting Assisted Culling, which starts the analysis in the background. That is the version worth knowing about, because it moves the work into the one stretch of the evening when you are already standing around waiting for a copy to finish.

In the cloud app, Lightroom, you open an album and choose Assisted Culling > Cull. Same three criteria and the same three reject rules, with slightly different wording on the controls: Classic says Subject Focus and Eye Focus, the cloud app says Subject Sharpness and Eyes Sharpness. If you are following a tutorial and the labels do not match your screen, that is why. We walk through both paths properly in our Assisted Culling setup guide.

The three things it looks for

Three, and only three. Each one is a measurement of the picture, not of the occasion.

The sliders are the part most people skip, and they are the whole feature. A threshold that is right for a controlled headshot set is wrong for a dim reception, because in a dim reception a good number of your keepers are softer than a studio frame and always will be. Run it per section of the day rather than once across the whole card.

The Select Can't tell images option under Eyes Open deserves a moment on its own. It decides what happens to frames where the model genuinely cannot say whether the eyes are open: sunglasses, a profile, hair across a face, a child mid-turn. Turn it on and those land in your shortlist for you to judge. Turn it off and they sit outside it, which is fine as long as you know that is what you chose.

The reject side, and what actually happens to your files

Adobe gives the reject side three rules, and you switch each one on or off:

• Exposure issues, which rejects photos with exposure problems.
• Misfires, which rejects photos captured accidentally: the frame that fired as the camera swung down, the floor, the inside of a bag.
• Documents and receipts, which rejects photos of documents and receipts.

That third rule reads like an odd thing to build until you remember what is actually on your cards. The shot list you photographed so you would not lose it, the seating chart, the printed timeline taped inside a door, the parking sign you shot so you could find the car at 1 AM. Every one of those is a frame you have to skip past manually on every single pass, forever, and now you do not.

Once the analysis is done, you can review Selects, Rejects or All, each marked with a badge, and then apply batch actions: a flag, a star rating, a color label, or adding the images to a collection or album. The cloud app's menu also offers Remove from Album, Copy to Cloud, and Delete.

Nothing at all happens to a file until you choose one of those, which is the right design. Be deliberate about which one you pick, though: a star rating or a color label is a note in a database and completely reversible on Tuesday morning, while Delete is not. This is also where our own approach differs by design. Kepla's picking step never deletes, moves or renames a file, so there is no version of the workflow where a mis-scored frame quietly leaves your drive. Everything it did not pick stays one tap away.

Read the scope statement before you trust it on a wedding

Two caveats come from Adobe rather than from us, and both are stated plainly. The first is that this is Early Access. The second is the one to write down: in Adobe's own words on its Assisted Culling walkthrough, "the current feature set is optimized for portraits and headshots."

That is a fair and useful thing for a company to say about a young feature, and it tells you exactly where to expect it to hold and where to expect it to wobble. Headshots, engagement portraits, school and corporate work, family formals: those are faces looking at a camera at a sensible shutter speed, which is the designed case. A first dance at 1/60th with deliberate drag, a backlit confetti exit, a documentary frame where the subject is one small figure in a doorway: those are not, and the scores you get back will read as though the shoot went worse than it did.

None of that is a defect. It is a scope, and knowing a scope is worth more than any star rating out of ten. For the full account of what this feature and the app around it still cannot do for you, we wrote that separately in Lightroom culling limitations, and it is the piece to read if your shoots look nothing like a headshot session.

A free way to find out how it does on your files

Any published accuracy figure for a tool in this category is either a vendor's own marketing or someone else's shoot, and your files are neither. The good news is that the honest test is free and takes an evening.

Pick a shoot you culled by hand several months ago, ideally one you no longer feel anything about, and one where you still have every original. Run Assisted Culling over the whole folder before you look at a single score, so your eye is not led. Then compare its Selects against the set you delivered, and study only the disagreements. Agreement teaches you nothing.

The disagreements come in two kinds, and they are not equally expensive. Frames it picked that you did not are cheap: you glance at them and move on, and a few seconds each is the whole cost. Frames you delivered that it put in the reject pile are the expensive kind, because in a normal week you would never have looked at them again. That second number is the one that decides whether you can trust it, and it is the number nobody publishes.

Repeat it on a second shoot of a different type, because a tool that agrees with you on formals can disagree with you all evening on a reception. If you want a fuller protocol, including how to count false rejects without kidding yourself, we set one out in how to test AI culling accuracy.

Where to put it in an actual evening

The mistake is to open the shortlist first and work only inside it. A shortlist looks finished, and this one is not: it is technically sound and creatively unsorted, which is a different thing.

A better order is to spend the first twenty minutes in grid view marking narrative anchors before any filter is applied. The reactions, the entrances, the toast, the thing that went wrong and turned out to be the best frame of the day. Then let the filters do what they are good at and thin the repeats and the obvious technical failures. Then go back to the unfiltered run around each anchor you marked, because the alternative expression almost always lives one or two frames either side of the one you flagged.

One practical note on speed. Analysis competes for the same machine that is rendering previews, so it feels slow if you start sorting the moment the import bar appears. Running it at import, or leaving it working while you verify card backups, costs you nothing you were using. If Lightroom feels like treacle generally, the cause is usually preview settings rather than this feature, and Smart Previews are the underused fix.

Who should use it, and who needs something else

Use it, without hesitation, if you shoot portraits, headshots, corporate or school work. That is Adobe's stated case, the feature costs nothing beyond the plan you already pay for, and the two minutes it takes to enable are the cheapest workflow experiment available to you this year. The same goes for anyone shooting a few hundred frames a session, and for screening a second shooter's card down to something you can look at properly.

Be more careful if you shoot high volume, burst heavy, low light work, or if your real problem is not analysis at all. A UK survey of more than 300 wedding photographers, reported by PetaPixel, put 55 percent of working time into editing against 11 percent into picking. If that shape matches your week, a faster first pass is a nice win and not the win.

And for a good number of photographers the bottleneck was never the software's speed. It is that none of this can begin until you are home, at the desk, at 11 PM, which is when your judgment is worst. Lightroom Classic runs on Windows and macOS and has never run on an iPad or iPhone, so the drive home stays unusable. We looked at which pass belongs where in phone versus computer culling, and compared the built in feature with the paid specialists in Lightroom culling versus dedicated tools.

That gap is the thing we are building for at Kepla: a first pass on the device already in your pocket, picks only, nothing ever deleted, moved or renamed, and the same shoot waiting on the Mac when you want a big screen. The picking app is still in development. The booking page is live and free today while we build it.

FAQ