Per-Image Detection: Why You Don't Pay for Photos Without Faces

There is a specific kind of billing worry that comes up constantly among pay-as-you-go photographers: will I get charged for every single photo I export, even the ones that do not actually need the tool I ran? It is a fair question, and the honest answer is no. Certain tools in Imagen detect real content on a per-image basis, and a tool that has nothing to act on in a given photo simply does not charge you for that photo. The clearest example is a face retouch: no detectable face, no charge.

What per-image detection actually checks for

Not every AI tool depends on specific content being present in a photo, but some clearly do. A face retouch needs a face. A tool built around skin or eyes needs the same. When one of these tools runs across a gallery, it is not applying a blind, one-size-fits-all pass to every frame regardless of content. It is checking what is actually in each photo and applying itself only where it makes sense to. Where there is nothing for the tool to act on, nothing happens to that photo, and nothing gets billed for it either.

Other content-dependent tools follow the same logic

Face retouching is the clearest way to explain this because almost every photographer has an intuitive sense of which frames in a shoot have a visible face and which do not. But the underlying principle is not specific to faces. Any tool built around detecting a particular kind of real content in a photo works the same way: it looks for that content first, and it only does billable work where the content is actually present. The practical takeaway is broader than one tool. Before you assume a per-photo charge applies uniformly across an entire export, it is worth checking whether the specific tool you ran is one that depends on detecting something, because that detail changes what you should expect on your invoice.

The face retouch example, in a real gallery

Most shoots are not wall-to-wall close-up portraits, even the ones built around people. A wedding gallery includes getting-ready details, an empty aisle before guests are seated, a table setting, a pair of shoes on a windowsill, alongside every portrait and candid of the couple and their guests. Run a face retouch tool across that entire gallery and it has real work to do on the portraits and candids, and nothing to do on the table setting or the empty aisle. Under per-image detection, you are not charged for a face retouch on the photos that never had a face in them to begin with.

Why this matters for fairness, not just cost

The value here is not only that it saves you money on any single export. It is that the billing model matches what actually happened to each photo. A flat per-photo charge for a specific tool, applied whether or not that tool did anything, would mean effectively subsidizing the frames it could not touch with the ones it could. Per-image detection removes that mismatch. You pay for edits that happened, on the photos where they happened, which is a fairly reasonable definition of fair billing for anyone paying per export.

The Imagen assistant finding and editing photos from a plain-language request

It also removes a small but real incentive problem. If every exported photo cost the same regardless of what a tool actually did to it, there would be a quiet financial reason to leave a face-dependent tool off a mixed gallery altogether, just to avoid paying for frames it could not help. Per-image detection means you can run the tools that make sense for a shoot without doing that kind of mental math first.

Which kinds of shoots benefit most

The more a shoot mixes people-heavy frames with everything else, the more this detail actually matters to your invoice. Think of a restaurant shoot: plated dishes, the dining room, the bar, alongside a handful of staff portraits for the website’s team page. Or a product launch event with keynote candids next to detail shots of the product itself on a table. In both cases, a face retouch tool has a clear, narrow job to do on a subset of the gallery, and per-image detection means you are not paying that tool’s price across frames it was never going to touch anyway.

A catalog shoot makes the pattern even clearer. Forty products photographed on a plain background, plus five lifestyle frames of a model wearing or using one of them for the homepage banner, is a gallery where a face-dependent retouch tool has exactly five photos to work with. Running that tool across the full forty-five costs you the same as running it across just those five, because the other forty were never going to be charged for a step they never needed.

How this interacts with exporting to Lightroom

Per-image detection is most visible when you export selectively and directly from Imagen, because your invoice tracks exactly which photos you chose to export with which tools applied. If you export a full album to continue in Lightroom the way you always have, you are exporting the entire edited gallery at once, so the selective, per-photo nature of this billing detail matters less in that moment. You are still only ever charged for tools that actually applied to a given photo, but the distinction is easiest to notice when you are choosing individual exports rather than sending an entire finished shoot onward.

The same logic extends to a shoot with almost no people in it at all. An architecture shoot or a still-life product catalog might contain zero frames with a detectable face across the entire set. Running a face retouch tool across a gallery like that costs nothing beyond the tool’s baseline check, since there is never a photo for it to actually retouch, and per-image detection means you are never billed for a step that had no work to do anywhere in the shoot.

A practical walkthrough

Say you shoot a corporate anniversary event: a keynote, audience candids, a step-and-repeat photo wall, and a series of product table shots for the marketing team. You run a face retouch pass across the whole gallery because most of it involves people, then export sixty selected photos to the client. Of those sixty, forty-five have visible faces and were genuinely retouched by the tool. The remaining fifteen are product table shots and wide venue shots with no detectable face. Under per-image detection, the face retouch charge applies only to the forty-five photos the tool actually worked on, not to all sixty just because you ran the tool once across the gallery. This is the kind of estimate that becomes easy to run mentally once you have seen it play out on a real invoice a time or two.

What this means for estimating a job’s cost upfront

Once you know a shoot’s rough content mix in advance, you can estimate the cost of a face-dependent tool before you ever run it. A corporate event with a printed agenda usually tells you roughly how many segments involve people versus signage, staging, or product displays. A restaurant or retail shoot usually splits cleanly into product-only frames and the smaller set that includes staff or customers. Neither estimate needs to be exact to be useful. Knowing that a face retouch pass will likely apply to a minority of a mostly product-focused gallery, rather than assuming it prices out the same as a portrait-heavy one, is often enough to quote a job with more confidence than guessing at a flat per-photo number.

Getting comfortable with this before you rely on it

If you have not paid close attention to your invoices before, it is worth testing this deliberately on one shoot: run a face-dependent tool across a genuinely mixed gallery, export a batch that includes both people and non-people frames, and compare what actually gets charged against what you expected. Seeing it once on your own numbers tends to settle the worry faster than any explanation can. For more on how specific tools behave across different kinds of shoots, the valuable tips library is a useful next stop.

What’s new in Imagen: The next step for Imagen is an intelligent, agentic Studio for professional photographers, one project where you cull, edit, review and export in any order. AI filters such as blur and closed eyes help you narrow a shoot before you decide what leaves the project, and you pay on export for the photos you choose to deliver. The Studio is rolling out gradually as a beta.

Frequently asked questions

Does per-image detection apply to every AI tool?

It applies to tools that depend on detecting specific real content, such as a face for a retouching pass. Tools that are not content-dependent in this way work differently.

Do I need to turn per-image detection on?

No. It is simply how the relevant tools bill by default. You do not need to configure anything for it to apply.

What happens if a photo has a partially visible or unclear face?

The tool is checking for a genuinely detectable face to act on. Frames without one are treated the same as any other photo the tool has no work to do on.

Does this change anything about the quality of the retouch itself?

No. It only affects billing on photos the tool never touched. Photos with a detectable face are retouched and charged exactly as they always would be.

Want to see per-image billing on a gallery of your own? Start a project in Imagen’s new Studio and export exactly what you need at imagen-ai.com.