Imagen AI review (2026): the culling side, honestly

Most reviews of Imagen talk about the editing, because that is the product Imagen is famous for. This one is about the culling, because that is what photographers keep asking us about. The short answer is that the culling is good, it is cheap once you are inside, and getting inside costs more than the sticker price suggests.

The short version

Imagen is an AI editing company first. You train it on your own edits, hand it a shoot, and it applies your style to every frame. That is the core product and it is strong. AI culling arrived later, and it is sold the way a good restaurant sells dessert: an add-on to a meal you already ordered.

The culling itself is a paid extra on top of an editing plan. It costs $12 a month billed annually, or $18 a month billed monthly, and that flat fee covers unlimited culling. You get two free culling projects to try before you commit. Everything runs in Imagen's cloud, which is the single most important fact in this review and the one that decides whether the tool fits your life.

So the honest framing is this. If you already want AI editing in your own style, Imagen is excellent and adding culling is close to free. If you only want help picking photos, you are buying an editing subscription to get to a culling feature, and your shoot has to travel to someone else's servers before the first decision gets made.

The price, including the part that is easy to miss

Imagen's published plan structure is clear about the dependency: you need an editing plan before you can subscribe to culling. There are two shapes of editing plan. Pay as you go is a monthly base fee plus a charge per photo you edit. Limitless is a flat fee that covers unlimited edits, unlimited AI tools, and culling included.

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

There is a detail here that works in Imagen's favor and that nobody mentions. Subscribing to culling waives the pay as you go monthly minimum, which their own culling page states plainly. So a culling subscriber is not paying $12 plus $7. The realistic floor for someone who wants culling and only occasional editing is $12 a month on annual billing, plus whatever editing they actually run. That is competitive with Aftershoot Select at $10 a month annually.

One inconsistency to know before checkout: Imagen's main pricing page currently lists culling "From $29/mo" while the culling page quotes $12 and $18. Check what your own account is offered rather than trusting either page, ours included.

What cloud based means on a Saturday night

Nothing happens until your photos are on Imagen's servers. That is not a criticism, it is the architecture, and it has consequences a desktop tool does not have.

A wedding is typically 2,000 to 4,000 RAW frames, which is roughly 60 to 120 GB on disk depending on your camera. In Imagen's favor, the app does not push full RAW files up the wire: their support docs confirm they compress photos before upload, similar to Lightroom's Smart Previews. That is a real saving. It is still an upload of an entire wedding, and Imagen does not publish what the compressed total looks like, so plan for substantial rather than instant.

Here is what a transfer of that scale means on ordinary home internet. This is arithmetic, not a benchmark.

Pure arithmetic at the stated speed. Real transfers run slower once overheads, throttling and a dropped connection at 2 AM are included.

Notice that upload speed, not download speed, is what matters, and on most home plans upload is the small number. On fiber this section is close to irrelevant. On a rural line, or hotel wifi at a destination wedding, it is the whole story.

The second issue is contractual rather than technical. Some clients, particularly corporate work, brand shoots, secure venues and anything under an NDA, restrict sending images to third party servers. That clause is often sitting in a contract you signed without reading closely. It does not make Imagen unusable, it makes Imagen unusable for that job, which is worse in a way because you find out late. We listed the clauses people miss in contract mistakes photographers make.

Where the cloud genuinely wins

Being fair about this matters, because the cloud model has real advantages that local tools cannot match and photographers talk themselves out of it too quickly.

Your computer does not matter. Imagen's own page makes the point that processing does not touch your CPU or GPU. A five year old laptop with a tired battery does the same job as a maxed out studio machine. For anyone putting off a computer upgrade, that is a genuine saving that belongs in the comparison.

Nothing melts, nothing blocks. Local AI culling pins your machine for the duration, so you cannot comfortably edit or export while it runs. With Imagen the work happens elsewhere and your machine stays yours. Twelve weddings in a month is Imagen's thermal problem, not yours.

One honest caveat on the "works from anywhere" pitch. Imagen's support documentation says that when you start a culling project you should complete it on the same computer. The heavy compute is in the cloud, but the project is still tied to the machine you began on. It is not the same as picking up a shoot on any device you happen to be holding.

How good is the picking?

Imagen's culling looks for the things every tool in this category looks for, and it names them clearly: blurry frames, closed eyes, and duplicates grouped by face recognition so the six versions of one moment appear together. It also has a nice touch for wedding work, a kiss recognition exception, because closed eyes during the kiss are the shot rather than the mistake. Anyone who has had a tool flag their best frame of the day will appreciate that.

The more interesting feature is Cull to Exact Number. You set a count or a percentage and the AI selects that many. If you sell a 600 image gallery, you can ask for 600. That is the sort of thing that comes from a company that thinks about delivery, not just detection.

Results arrive as star ratings inside a review interface, split into keepers, standalones and duplicates. As with every tool in this category, you are auditing a proposal rather than being handed a finished gallery, and the useful question is not whether it is right but whether its mistakes are the kind you can catch quickly. Our take on how far to trust any of these models is in can you trust AI to pick your photos.

Test it properly with those two free projects. Run a wedding you already culled by hand, then look only at the frames where you and the machine disagreed. That exercise tells you more than any review, and it costs nothing.

How it compares to the culling specialists

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

Read that table by the second column rather than the third. The prices are within a few dollars of each other and none of them will decide your year. Where the work happens will. Aftershoot and Narrative keep the shoot on your drive and start the moment you plug in the card, at the cost of your machine doing the work. Imagen starts after the upload, and costs your machine nothing. Photo Mechanic decides nothing at all and simply makes you faster at deciding, which we covered in the best photo culling software.

If your real bottleneck is editing rather than selection, the calculus flips entirely and Imagen becomes the obvious buy. A UK survey of more than 300 wedding photographers, from 2020 and so before these tools were common, found editing eating 55 percent of working time against 11 percent for culling. If that split matches yours, buying a culling specialist is optimizing the small number.

Where Kepla sits, said plainly

We are building a competing tool, so treat this section with the skepticism it deserves. Here is the difference in one sentence: Imagen starts when the upload finishes, and Kepla starts in the car.

Kepla is native on iPhone, iPad and Mac. The first pass happens on the phone on the drive home, with no upload to wait for and no desk to sit down at. When the shoot needs real scrutiny, the same job opens on the Mac and you keep going: two near identical frames side by side, zoomed in, everything still there. One shoot, two screens, nothing done twice. The argument for that shape of workflow is in phone versus computer culling.

Kepla clears. It never deletes, moves or renames a file. It marks which photos it chose, and from 2,000 frames you get roughly 800 picks in minutes. Everything it did not pick is still exactly where you left it, one tap away.

And it books the job. None of the tools in the table above will send a client a booking page, hold a date or take a deposit. Kepla's booking page is live and free today.

Who should buy it, and who should not

Buy Imagen if

• You want AI editing in your own style. This is the real product and it is very good at it. Culling at $12 a month on top is one of the better add-on deals in this category.
• Editing is your bottleneck, not selection. If your backlog is color and exposure rather than choosing, nothing else on the list touches the problem.
• Your computer is old or busy. Cloud processing means the machine you have is the machine you need.
• You have real upload speed. Fiber makes the biggest objection in this review disappear.

Look elsewhere if

• You only want culling. You are buying an editing subscription to reach a culling feature. A specialist costs about the same and starts immediately.
• Your upload is slow. A first pass you cannot start until tomorrow afternoon is not a time saving. The options are laid out in Imagen alternatives.
• Your contracts restrict third party servers. That is a hard stop for those jobs, whatever the price is.
• You want the first pass done before you get home. Nothing mature exists for that today.

Imagen is a good tool built by people who understand post production better than most. The criticism here is not that the culling is weak, it is that culling is the second thing the company does, and it is priced, packaged and architected accordingly. Know which of the two products you are actually buying and you will be fine. To see the whole field side by side, our comparison page lays it out.

FAQ