Home · Blog · Culling & AI

Culling on your phone vs your computer: which for which job

Osmel Contreras · Founder, Kepla · August 2, 2026 · 8 min read
A woman in a red dress seated on rocks by the sea, looking away, a near duplicate frame
Culling & AI

This gets argued as a loyalty test: real photographers cull at a desk, everyone else is cutting corners. That is the wrong frame. Culling is not one task, it is a stack of different judgements, and they have wildly different hardware requirements. Some of them are fine on a phone. Some of them are impossible. Here is the split, made honestly, and the workflow it points to.

01 · THE FALSE CHOICE

Culling is not one job

Open any shoot and look at the actual decisions you are making. A wedding at 2,000 to 4,000 RAW frames is not 3,000 equally hard calls. It is a few hundred hard ones buried under a couple of thousand easy ones.

The easy ones are the eyes-shut frame, the one where the flash did not fire, the six identical shots of the same handshake, the frame where a waiter walked through. You do not need a monitor to make those calls. You need about a second per frame and a screen you can see.

The hard ones are different in kind. Which of these two nearly identical portraits is sharper on the eye. Whether this frame is soft enough to matter at print size. Whether the skin tone in the reception shot is going to fight the rest of the set. Those calls need pixels, comparison and control.

Once you separate those two piles, the phone versus computer argument dissolves. It is not a question of which device culls better. It is a question of which decisions each device is built for.

02 · WHAT A PHONE IS GOOD AT

The small screen's real strengths

Obvious rejects

Gross failures announce themselves at any size. Motion blur, a completely missed focus, a closed eye, a frame of the floor. A phone is perfectly capable of telling you those are gone, and getting them out of the way early makes everything after it easier, because you stop judging your good work against a backdrop of wreckage.

Expression

This is the surprising one. Faces are the thing human vision is most specialised for, and reading a smile does not require resolution. If anything, a small screen helps: you see the frame as a whole, the way a viewer scrolling a gallery will, rather than getting lost in an eyelash. When you are choosing between four frames of the same toast, the phone gives you the right kind of look.

Story and rhythm

Which moments made the day, which parts of the evening are over-covered, whether you have the grandmother at all. These are questions about the set, not about any frame, and they answer fine at thumbnail size.

Time you already have

The biggest advantage is not perceptual, it is practical. The drive home, the hour on the sofa, the wait at the airport. A UK survey of 300 plus wedding photographers found the working week runs about 4 percent shooting, 55 percent editing and 11 percent culling. Culling is the gate: nothing downstream starts until it is done. Moving the first pass into dead time pulls the whole schedule forward without adding a single working hour.

03 · WHAT A PHONE IS BAD AT

The limits, stated without hedging

Focus checks at 100 percent

You cannot reliably tell sharp from slightly soft on a phone, and pinch zoom does not fix it. You get a magnified view of a preview, held in a moving hand, with no stable reference. If a frame's fate turns on critical focus, that decision has to wait for a proper display. This is the single clearest ceiling on mobile culling and it is not going away.

Near-identical frames

Choosing between six versions of the same group shot means holding six images in mind at once, and human working memory is bad at that. Desktop tools solve it by putting the frames side by side at size. A phone shows you one at a time and asks you to remember, which is exactly the wrong cognitive load. This is a big part of why the job feels harder than it should, as we covered in why picking the right photo is so hard.

Colour and exposure

Phone screens are bright, contrasty and enthusiastically processed. That is great for looking at photos and bad for judging them. Anything to do with white balance consistency, shadow recovery or whether a frame is rescuable belongs on a screen you have profiled.

Raw throughput

A thumb is slower than a keyboard. If you are moving through thousands of frames making binary calls, keys win, which is why the culling shortcuts worth learning are still worth learning.

04 · JOB BY JOB

Where each decision belongs

DecisionPhoneComputerWhy
Obvious rejectsYesWorks, but wastefulVisible at any size, no reason to wait
Best expression of fourYesAlso fineFaces read well small; whole-frame view helps
Is this moment coveredYesAlso fineA question about the set, not the frame
Sharp or slightly softNoYesNeeds a stable 100 percent view
Two near-identical framesNoYesNeeds side by side at size
Colour and exposure callsNoYesNeeds a profiled display
Album shortlistPartlyYesSequence and crop need room
Sneak peek for tonightYesOverkillSix frames, fast, from anywhere

Notice which column has more rows. The phone handles most of the volume, the computer handles most of the difficulty. That is a good division of labour, not a compromise.

05 · THE WORKFLOW

First pass, then detailed pass

Here is the model, in five steps, with no software required to believe it.

  1. Back up first. Two copies before a single decision. Nothing in this workflow is destructive, but the rule stands anyway. See the backup workflow.
  2. First pass on the phone. Binary only. Obvious out, obvious in, everything uncertain stays where it is. Do not zoom, do not deliberate, do not delete. Speed is the point.
  3. Arrive at the desk with the pile already sorted. Instead of 3,000 unopened frames you have a shortlist and a remainder, both intact.
  4. Detailed pass on the computer. Side by side on the near-duplicates, 100 percent on the focus questions, one pick per moment. This is the pass that decides the gallery.
  5. Shortlist tier last. Album spreads and portfolio candidates, chosen from picks only. Binary first, stars later, as argued in star ratings vs flags.

The gain is not that a phone is faster per frame. It is that the first pass happens at all, in hours that were otherwise spent on a podcast, and the desk session starts with the easy 70 percent already handled. The full desktop method is in how to cull 3,000 wedding photos in under two hours.

The one thing that can break this

A two-device workflow only pays if the handover is clean. The test is simple: open the shoot on the computer and check whether the picks you made on the phone are already marked on the original files, with nothing to match by hand. If they are not, you have not saved an hour, you have spent one twice. This is why improvised phone culling usually gets abandoned after two shoots, and it is the part any serious mobile tool has to solve before anything else.

06 · WHAT THIS MEANS FOR TOOLS

Why nobody sells this yet

The tools that exist were built for one screen. Aftershoot and Narrative Select are good desktop programs and they process locally, which many photographers prefer. Imagen lives in the cloud but wants the whole shoot uploaded before it does anything. All of them assume you are already sitting down. None of them has a first-pass story, because none of them has a phone.

That is the gap Kepla is being built into, and the argument above is the design brief, not the marketing. Kepla clears the obvious misses on your Mac: every frame checked against clear criteria, sharp focus, open eyes, expression, near-identical frames, and a shortlist handed back to you. From a 2,000 photo wedding that is roughly 800 picks, in minutes. You approve on the small screen, in dead time.

Then the same shoot opens on the Mac and you keep going: side by side, zoomed in, one shoot across both screens with nothing redone. Every frame it did not pick is still there, because it never deletes, moves or renames a single file. It marks what it picked, and anything else is one tap away. The apps are still being built, and we will say so until they ship. The booking page is live and free today, the comparison page shows where everything stands, and the cost calculator will tell you what your current culling hours cost.

07 · COMMON QUESTIONS

FAQ

Is culling on a phone accurate enough to trust?

For the decisions a first pass is made of, yes. Blinks, badly missed focus, someone walking through the frame and which of six similar shots has the best expression are all readable on a phone. What a phone cannot settle is fine focus at 100 percent, small differences between near-identical frames, and colour. Keep those for the desktop and a phone pass is reliable.

What is the fastest way to cull a big shoot?

Two passes, not one. A quick binary sweep that removes the obvious failures and marks the obvious keepers, then a slower pass on what is left where you compare, zoom and choose one frame per moment. Doing both at once is what makes culling take an evening, because you are asking a hard question of every frame including the ones that were never in contention.

Should I stop culling on my computer entirely?

No. The computer is where the hard half of the job belongs: side by side comparison, 100 percent focus checks, colour judgement on a calibrated screen and keyboard speed. The phone is not a replacement for that, it is a way to arrive at it with most of the volume already sorted.

Does a two-device workflow mean doing the work twice?

Only if your tools make you. The test for any setup is whether the picks you made on the phone are already marked on the original files when the shoot opens on the computer. If you have to re-do decisions or match filenames by hand, the workflow is broken and you should fix the handover before you change anything else.

FOUNDING COHORT · 100 SEATS

Never cull alone at 1 AM again.

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.

Request Founding Access
KEEP READING