BRENNEN

Industries / Retail, hospitality and personal services

Photography

Photography as a trade takes pictures to order for people and businesses, such as portraits, weddings and product shots, and sells pictures already taken, together with the editing, printing and delivery that follow.

1839

On 19 August 1839 Louis Daguerre's process, whose rights the French state had acquired in exchange for a lifetime pension, was explained step by step before a joint session of the Académie des Sciences and the Académie des Beaux-Arts in Paris, so that anyone could learn it.

metmuseum.org

40

Milestones achieved across 7 domains.

8

To watch for: announced, not yet confirmed outside.

15

Open deliverables. Nobody has shipped them.

In one minute

Where it started

Daguerre's process was made public in Paris in 1839, and by March 1841 Richard Beard had opened what the Victoria and Albert Museum calls Europe's first public photographic studio, in London. From 1854 the carte-de-visite gave many cheap prints from one sitting, and from 1888 Kodak sold developing and printing as a service to anyone with a camera, by its own account.

Where it stands

Almost everyone now takes pictures: Photutorial, a photography website, estimates that about 2.1 trillion photos were taken in 2025, and that 94% were taken on smartphones in 2024. The paid trade is made of many small firms: Eurostat counts 201,479 photography businesses in the European Union in 2024, up from 165,004 in 2021, with a combined turnover it estimates at about €10 billion, an average of about €50,000 each. Software now does part of the work: by IKEA's account most of its single product images were computer-generated by 2014, Zalando says more than 90% of its content is made with AI, and Shutterstock sells its archive to makers of AI models.

What is open

Most claims about the tools the trade now uses come from their sellers, and nobody else has measured them: how faithfully AI tools show a garment or a customer's face, whether retouching shifts skin colour, how often selfie search shows a guest someone else's photos, how labs crop and colour prints, and which services strip a photographer's name and rights from the files. Each can be tested with consenting adult volunteers, test files and free software. AI models that read pictures and documents now let a small team score thousands of photos, contracts or claims.

Domains

AchievedWatch forOpen

Advances to build with

3 research notes

Open in photography

3 of 15 open deliverables

Open · Portraits

Passport photo apps, tested

A public test of the phone apps and shop software that sell passport photos with an automatic rule check. Adult volunteers who agree to take part are photographed under a fixed method, once in a correct pose and again with one planted fault each (mouth open, glasses, a shadow on the face, a tilted head, a busy background). Every photo goes through every app. The test counts how often each app passes a faulty photo and how often it rejects a good one, for each skin-tone group. No photo is sent to any government system.

Built with what exists

OFIQ, the open-source face image quality software from Germany's Federal Office for Information Security (BSI) that implements ISO/IEC 29794-5, as a reference checker; the compliance model trained on DFIC, from its GitHub page, as a research baseline; commercial passport photo apps such as PhotoAiD; Google's Monk Skin Tone Scale for labeling skin tone; a fixed light and a plain backdrop.

Credit · unclaimed

Open · Weddings and events

Selfie search, tested on volunteers

A staged event, such as a mock reception or a 5 km fun run, where every person photographed has signed up for the test, followed by a test of commercial selfie-search galleries on its photos. Every face in every photo is labeled by hand. For each service, the test measures the share of a volunteer's photos it finds and the share of photos it shows to the wrong volunteer, which is a privacy leak as well as a miss.

Built with what exists

Trial accounts on commercial gallery services with selfie search, such as Kamero and Pic-Time; InsightFace's open face models as a baseline; CVAT or Label Studio for labeling faces; signed consent forms that cover the test and its deletion date.

Credit · unclaimed

Open · Commercial work

AI garment photos, measured

A public test of how faithfully AI tools show a real garment when they put it on a generated model or a new background. Measure 100 garments at marked spots with a spectrophotometer, an instrument that reads a colour as numbers. Photograph each garment flat and on a consenting model under controlled light, with a colour chart in the frame. Give the flat photos to several AI on-model tools, then measure the colour error of each result against the garment itself in CIEDE2000 (ΔE00), the standard measure of how different two colours look, and have two people count changed details such as buttons, pockets, seams, prints and logos.

Built with what exists

A handheld spectrophotometer such as an X-Rite Ci64 or a Nix Spectro 2; a Calibrite ColorChecker chart; a camera and continuous lights; open try-on models such as IDM-VTON or CatVTON from Hugging Face, and trial accounts of commercial on-model tools; Python with the open colour-science library to compute ΔE00.

Credit · unclaimed