Photoroom and the surprisingly profitable art of looking real

The trouble began when the clothes seemed to float. Depop, the secondhand fashion marketplace, had tried giving sellers a cleaner background. The editing worked. The business result was less obliging. According to Photoroom’s account of the tests, only 9% of eligible users posted with the background removed, and listing and sales metrics were flat. Then the team added AI-generated shadows. Items listed rose 1.5%. A small patch of darkness had given the merchandise its weight back.

• Photoroom turns ordinary product photos into listing and campaign images.
• Sellers use the app; larger businesses automate through its API.
• The useful lesson: measure completed listings, and check that the image still tells the truth.

There is something pleasingly awkward about this result. A technology company removes the messy world from a photograph, then discovers it must put a little of the world back. Buyers need evidence that an object exists. A shadow is one modest piece of that evidence. The lesson travels beyond photography: removing friction can also remove the cues that make an experience familiar.

The lasso was the clue

Photoroom’s origin is another small indignity. Its company history describes Matthieu Rouif, then a product manager at GoPro, spending hours tracing edges with a lasso tool while a designer was away. The computer could do extraordinary things, yet somebody still had to draw around an object. Rouif and engineer Eliot Andres founded Photoroom in 2019. The annoying task became the entry point.

That choice matters. Background removal has an intelligible before and after. A seller photographs a shoe, removes the room behind it and gets an image suitable for a listing. There is no need to develop a taste for AI. There is a job waiting to be finished, and the seller can judge whether the software helped.

Today the offering extends into generated backgrounds, retouching, shadows, resizing, virtual models and video. Batch editing repeats a treatment across a collection. Mobile and web access let the same work move between devices. For developers, the API takes an image, applies requested changes and returns an edited image, as its documentation explains. The glamorous vocabulary conceals a practical ambition: fewer chores between taking a photograph and publishing it.

A spreadsheet with a studio attached

At Decathlon, the chore had acquired industrial proportions. The sports retailer’s published case study describes agency edits taking two weeks per 1,000 images, with 15,000 waiting. Photoroom connected to the digital asset system through a Google Sheet. Presets assigned the desired treatment; edited files returned with the right names. The reported processing time for 1,000 images fell to 20 minutes, and per-image editing costs fell 99%.

“The first objective of a packshot picture is to let the customer understand the product.”Thomas Schouteeten / Product Content Leader, Decathlon

Those are vendor-published customer results, rather than promises for every catalog. Still, the spreadsheet is instructive. Automation needs an entrance people understand and an exit the existing system accepts. A dazzling image generator that leaves staff renaming files has merely moved the work to a less photogenic corner of the office.

Wolt illustrates a complementary arrangement. Its June 2026 account describes high-volume API processing alongside a web editor for local teams across 27 countries. Keeping an accessible editing tool beside the automated pipeline makes organizational sense. A merchant’s peculiar photograph does not become ordinary because the software expects it to.

Two cents, plus the work around it

Photoroom sells subscriptions to individuals and teams, API usage to software builders and negotiated plans to enterprises. The public pricing page, checked October 4, 2026, displays Pro at $7.50 a month when billed annually, Max at $20.99 and Ultra at $82.50. These are annual-billing equivalents; regional prices and allowances vary. Its API FAQ lists Basic calls at $0.02 and Plus calls at $0.10.

Processing 10,000 images once at those API rates implies $200 or $1,000 in call fees. Integration, retries and human review belong in a separate budget. That distinction is useful for a buyer: calculate the cost of an accepted, published image. A cheap output that needs a long inspection can become an expensive bargain.

Investors financed the expansion. Balderton led a $19 million Series A in 2022; the $43 million Series B in February 2024 brought announced total funding to $64 million. Sifted profiled the company as profitable that October. Profitability gives the story substance, although a historical report cannot settle today’s margins.

Paris was a starting point

The founders treated distribution as part of the product. In a 2023 account, Rouif says Photoroom launched in 15 languages and reached 28 within three months. Its first app-store feature came from Vietnam; its first press coverage came from Japan. English became the working language. That is a more concrete expression of international ambition than putting a globe on a slide.

The product must survive the picture

Its position sits between general creative editors such as Photoshop and Canva, focused photo tools such as Pixelcut, and the agencies doing repetitive retouching. Photoroom’s commercial argument is specialization: product photos, repeated rules and connections to selling systems. A studio remains useful when a shoot requires physical demonstrations, exact material capture or creative direction that a source image cannot supply.

The harder boundary is accuracy. In its August 2026 update, Photoroom describes an AI edit inventing stripes on a jacket sleeve. An earlier quality model missed it; an upgraded check flagged it. This is an unusually revealing example. A plausible image can contain a fictional product.

For a seller, the sensible starting point is a clear original photo, a repeatable treatment and a small test against existing results. Inspect logos, seams, color and dimensions before expanding the batch. Photoroom can shorten the distance to a finished listing. The shadow lesson suggests the right standard: let the software improve the presentation while the object keeps its identity.