Kepla vs Photo Mechanic: two different jobs

Two honest caveats before anything else. Kepla's picking app is still being built, so you cannot buy it today. And Photo Mechanic does not decide anything about your photos, which means this is not really a comparison of two competitors. It is a comparison of two different jobs, and working out which job is costing you your evenings is the actual decision.

What Photo Mechanic does today

Photo Mechanic, made by Camera Bits, is a photo browser and an ingest tool. It is not an editor, it does not develop RAW files, and it has no AI selection of any kind. Nothing inside it forms an opinion about which of your photos are good.

What it does instead is put a frame on screen the instant you press the arrow key. Your camera already writes a full size JPEG preview inside every RAW file. Photo Mechanic reads that preview rather than rendering the RAW, so there is no import, no catalog to build and no progress bar. Card goes in, photos are on screen. One architectural decision, twenty plus years of reputation.

The half of the license nobody talks about is the infrastructure around the looking. It copies from a card to two destinations at once, renames every file from a metadata pattern, and stamps caption and copyright on during the copy. The stationery pad applies IPTC data to a whole shoot in one move. Code replacements turn a jersey number into a full name. For press, sports and editorial work, that is not a nice extra, it is the deliverable.

We covered the software properly in our Photo Mechanic review, and the head to head with Lightroom is in Photo Mechanic vs Lightroom for culling. This page is about a different question: how do you weigh a mature, shipped tool against one that has not shipped, when the two are not even trying to do the same thing?

What it costs, checked today

Every figure below came off Camera Bits' own purchase page on August 12, 2026.

Prices checked August 2026 from each vendor's own pricing page.

Read the perpetual line carefully, because it is the one people misremember. Camera Bits states that a perpetual license includes one year of updates and support, after which you keep running the version you own on compatible operating systems. You are buying a version, not a subscription to the future. That is a genuinely good deal if the software already does what you need, and a slow disappointment if you expect it to keep pace with a new operating system in four years.

Worth naming plainly: $149 a year is more than a year of most AI selection tools. That is a strange sentence to write about a tool that used to be the cheap alternative to a subscription, and it is not an argument against buying it. It just means the money question and the workflow question are now the same question.

Why this is not a head to head

Here is the framing that makes the whole comparison stop being confusing. Photo Mechanic and AI selection tools answer two different questions.

Photo Mechanic answers how fast can I look at 3,000 photos. It closes the gap between one decision and the next, and it stays fast at the end of a long night when your judgment is fading and the last thing you need is a reason to lose your place. On a wedding of 2,000 to 4,000 RAW frames, half a second of render lag per photo is twenty to thirty minutes of staring at nothing.

Kepla is being built to answer which 800 of these 3,000 should I look at. Different question entirely. One shortens the road, the other shortens the trip.

The honest limit of speed is that it does not reduce the number of decisions. If you agonize over near identical frames, a faster viewer gets you to the agonizing sooner. It is a faster road, not a shorter one. The honest limit of selection is that it does not do the looking for you. You still review a shortlist, you still overrule it, and you still sign off. Neither category has ever finished a gallery for anybody.

Which of those two problems is actually eating your week is the entire buying decision, and most photographers have never measured it. Take one real job and write down two numbers: how long you spent waiting for photos to appear, and how long you spent deciding. If the waiting number is big, buy Photo Mechanic and stop reading. If the deciding number dwarfs it, no viewer on earth fixes that. How long culling actually takes has a method for measuring both without kidding yourself.

And there is no rule saying you pick one. Ingest in Photo Mechanic, let a selection tool propose a shortlist, then review that shortlist at full speed. Reviewing a proposal is still looking at photos, and looking at photos is the thing Photo Mechanic is best in the world at.

Who should buy Photo Mechanic this month

If you need something working on the next job in your calendar, buy the shipped tool. Ours is not finished, and telling you to wait would be a bad trade for you and a cheap one for us.

• You ingest from cards constantly. The copy, the rename, the two destination backup and the baseline metadata in one pass is worth the price on its own, before you look at a single frame.
• Metadata is part of what you deliver. Press, editorial, agency, anything with a caption standard. Nothing else handles IPTC this well.
• Your deadline is measured in minutes. Sports and news. This is the job the tool was built around and it is not close.
• Your computer is not new. Photo Mechanic decodes nothing, so its speed barely moves with the age of your machine. AI tools have to analyze every frame, which is real work for an older processor.
• You already have the muscle memory. A workflow you can run without thinking is worth more than the license, and rebuilding one costs weeks. Pair it with a proper keyboard flow and your hands move at the speed of your eye: keyboard shortcuts for culling covers that.

Skip it if your problem is the sheer volume of decisions rather than the speed of the viewer, if you shoot a handful of jobs a year, or if you want one application instead of two. Photo Mechanic is always a second app. You still need an editor after it, forever.

The different bet we are making

Full disclosure: we are building a tool in the other category. Read this as a description of a design choice, not a review of something you can install.

Photo Mechanic assumes the work happens at a desk after you get home. In 1998 there was nowhere else for it to happen, and that assumption has quietly survived into every desktop tool since, AI ones included. Our bet is that the assumption is the problem. Kepla is being built to make the first pass on your iPhone or iPad, in the passenger seat on the drive home or on the couch the next morning, and to open the same shoot on your Mac when a job needs two nearly identical frames side by side at full size. One shoot, both screens, nothing done twice. That argument is set out in full in phone versus computer culling.

Second difference, and it cuts directly against an ingest tool's philosophy. Kepla marks which photos it picked and does nothing else to your files. Nothing is deleted, moved or renamed. Photo Mechanic renames everything on the way in, by design, and for a photographer with a naming convention that is a feature rather than a flaw. Both stances are defensible. They suit different temperaments, and it is worth knowing which one you are before you commit to a workflow.

Third, Kepla also books the job. Photo Mechanic does one thing extremely well and makes no claim beyond it, which is honorable. Kepla is being built as the whole back office, and the piece that works today is the booking page: live, free while we build, and the only part we will call finished.

What we refuse to claim until we ship

Three sentences you will not find on this site, and why.

We will not say we are faster than Photo Mechanic. We would lose that race on its own terms anyway. Nothing beats reading a JPEG the camera already made, and a tool that analyzes every frame is doing more work by definition. The time we are trying to save is on the other side of the ledger, in the number of frames you look at, and that is not measurable until you can run it yourself.

We will not publish an accuracy claim. Software can objectively check whether eyes are open and whether focus landed. It cannot check whether a frame was the moment. Any single percentage mixes both and quietly scores only the easy half. Several tools in this category publish those numbers from their own testing on their own galleries. When we eventually publish anything, hold it to exactly the same scepticism, because it will have exactly the same problem.

We will not imply the app is available. It is still being built, native for iPhone, iPad and Mac, and that sentence stays on the page until it stops being true.

There is also a whole half of Photo Mechanic we are not competing with at all, and pretending otherwise would be dishonest. Ingest to two destinations, filename templates, IPTC stationery, code replacements, contact sheets, direct upload to an editor's server. None of that is in our scope. If those are the parts you rely on, no AI selection tool replaces them, ours included. What people actually trust these models with is worth reading up on in do photographers trust AI culling.

How to judge us when the app ships

When it does ship, do not take our word for anything. Here is the test we would like to be measured by, and the useful part is that Photo Mechanic gives you a fair baseline to measure against.

1. Time your current pass first. One real job, in whatever you use now, with a stopwatch. Note the waiting time and the deciding time separately. Without that baseline every new tool feels fast.
2. Use a gallery you already delivered. Six or twelve months old, with the delivered set still in a folder. That set is your answer key, scored against your taste rather than a vendor's.
3. Count exactly two numbers. How many of your real delivered picks did it find, and how many extra frames did it put in front of you that you would never have kept.
4. Look only at the disagreements. The frames it missed are usually the emotionally loaded, technically imperfect ones, and they tell you what it cannot see. The frames it flagged that you passed on tell you how long you will still be sitting there at 11pm.
5. Check what happened to your files. Open the original folder afterward. Every frame should be exactly where you left it, named exactly as it was. If any tool fails that check, including ours, that is a reason to walk away.

The full method is in how to test AI culling accuracy, and the same test works on the shipped tools today. We make the same offer against the other end of the market in Kepla vs FilterPixel.

Until then, the fair summary is short. Photo Mechanic is the fastest way to look at photos, it has been for twenty years, and if looking is your bottleneck it is still the right purchase. If the number of decisions is your bottleneck, buy something in the other category this season and come and test us when we are real.

FAQ