BRENNEN

Industries / Retail, hospitality and personal services / Photography

Portraits

This work makes posed pictures of people for them to keep, give or use, from studio sittings and school picture days to passport photos and business headshots.

6

Milestones achieved, from 1841 to 2020.

1

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1841

    Portraits sold to the public

    Richard Beard, a London businessman, bought into and patented Alexander Wolcott and John Johnson's camera in 1840, and in March 1841 opened what the Victoria and Albert Museum calls Europe's first public photographic studio, at the Royal Polytechnic Institution on Regent Street. In July 1841 he bought the sole patent rights to the daguerreotype in England and Wales and licensed others to use it, on condition that they stamp their pictures "Beard patentee". The museum finds scant evidence that he spent much time behind the camera himself.

    LessonFrom the first studio, the person who sold the portrait and the person who took it could be different people.

    api.vam.ac.ukcollection.sciencemuseumgroup.org.uk

    Credit

    Richard Beard; Alexander Wolcott and John Johnson, whose camera he used

  2. 1854

    Many portraits from one plate

    André Adolphe-Eugène Disdéri found a way to expose several images on one wet collodion glass negative and print them together on albumen paper, then mounted each print on a card the size of a visiting card. He patented the carte-de-visite in 1854. Because one sitting now gave many cheap prints, most people could buy cards of famous sitters or order cards of themselves to give to friends. By the Royal Collection Trust's account, more than 300 million cartes were sold in England alone in 1861.

    rct.uk

    Credit

    André Adolphe-Eugène Disdéri

  3. 1925

    A booth takes the picture

    Anatol Josepho's Photomaton, a coin-operated booth that photographed the sitter and developed the prints without a photographer, opened near Times Square in New York in September 1925. It charged 25 cents for eight photos, which took about 8 to 10 minutes to print, and about 280,000 people used it in its first six months, by the J. Paul Getty Museum's account. In 1927 and 1928 a new Photomaton Company paid Josepho $1 million, plus a promise of royalties, to place booths across the country.

    getty.edu

    Credit

    Anatol Josepho

  4. 2013

    Store portrait studios shut

    CPI Corp., which ran more than 2,000 portrait studios inside Sears, Walmart and other US stores, closed all of them with little warning in April 2013, after its lenders' deadline passed. In the first three quarters of its last fiscal year its loss had grown fivefold to more than $60 million and its sales had fallen 24% to $192.7 million, by The Wall Street Journal's figures, and NPR reported that digital photography had cut into its business. CPI said it was "attempting to fulfill as many customer orders as possible."

    LessonA chain built on selling prints from cheap in-store sittings could not survive once families took and shared most of their own pictures.

    npr.org

    Credit

    CPI Corp

  5. 2018

    Picture day joins Shutterfly

    On 2 April 2018 Shutterfly, an online photo-book and print company, bought Lifetouch, which Shutterfly calls "the national leader in school photography", for $825.0 million in cash. Lifetouch also ran more than 400 studios under the JCPenney Portraits name and photographed church directories, and it brought in 39% of Shutterfly's 2018 revenue from the date of purchase. Shutterfly's annual report notes that parents could increasingly order and receive their child's school pictures online instead of through the school.

    sec.gov

    Credit

    Shutterfly; Lifetouch

  6. 2020

    Photo checker fails darker skin

    The UK passport website checks each uploaded photo automatically against rules such as a neutral expression and a closed mouth. BBC News fed more than 1,000 photographs of politicians into the checker and found that it called photos of dark-skinned women poor quality 22% of the time, against 14% for light-skinned women, and 15% for dark-skinned men against 9% for light-skinned men. A Black student, Elaine Owusu, was told her mouth looked open on each of five photos. The Home Office said more than nine million people had used the service.

    LessonThe photo rules were the same for everyone, but the software that enforced them was not equally accurate for everyone.

    bbc.com

    Credit

    Maryam Ahmed, BBC News, who ran the test; the UK Home Office, which runs the checker

Watch for

  1. Passport photos from a phone

    The US Secretary of State, Marco Rubio, announced that Americans will be able to take or upload their own passport photo on a personal device when applying through the State Department's new America.gov site. Software would check the photo as it is taken, for expression, centring, shadows, and items such as glasses and hats. The State Department says poor photos are the most common reason applications are held, and the report cites its figure of more than 300,000 applications declined over photos in 2024. Until now most applicants paid a pharmacy or photo shop about $12 to $18 for printed photos, the report says.

    Achieved whenThe upload is open to the public, and an outside party publishes how often the automatic check passes photos that are later rejected and rejects photos that meet the rules, for people of different skin tones.

    thetravel.com

    Claimed by

    US Department of State, September 2026

Open

  1. 01

    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.

    The brief
    Credit · unclaimed
  2. 02

    AI headshots checked for likeness

    A public test of whether paid AI headshot services return pictures that still look like the customer. Adult volunteers each give a service the selfies it asks for. Each output is scored two ways: by an open face recognition model, which ranks it against a gallery of real photos of all the volunteers and checks whether the right person comes first, and by people who know the volunteer, who try to pick their friend's generated headshot out of a lineup of generated headshots of other volunteers.

    The brief
    Credit · unclaimed

Open 01 · the brief

Passport photo apps, tested

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

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.

Why it is still open

BBC News tested one government checker in 2020, with politicians' photos. Wassim Kabbani, Kiran Raja, Raghavendra Ramachandra and Christoph Busch (2025) tested the quality measures in the ISO/IEC 29794-5 standard across age, gender and skin tone and found two that varied considerably with skin tone, but they tested the standard's measures, not the apps people pay for. Nuno Gonçalves and colleagues (2026) released DFIC, about 58,000 labeled images of more than 1,000 people for training compliance checkers, but its terms allow non-commercial research only and forbid passing the images on, so it cannot be fed to commercial apps. This research found no test of the commercial apps.

bbc.comarxiv.orgarxiv.orgraw.githubusercontent.com

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.

  1. First step

    Photograph ten consenting colleagues under the method and run their photos through two apps and OFIQ.

  2. Done looks like

    At least 100 consenting adults spread across the Monk scale, 10 photos each (two correct, eight with one planted fault), run through at least four apps and OFIQ, with each app's false-pass and false-reject rates by skin-tone group and their 95% confidence intervals published with the method.

Open 02 · the brief

AI headshots checked for likeness

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public test of whether paid AI headshot services return pictures that still look like the customer. Adult volunteers each give a service the selfies it asks for. Each output is scored two ways: by an open face recognition model, which ranks it against a gallery of real photos of all the volunteers and checks whether the right person comes first, and by people who know the volunteer, who try to pick their friend's generated headshot out of a lineup of generated headshots of other volunteers.

Why it is still open

Ringover's 2024 survey of 1,087 recruiters measured preference (76.5% preferred AI headshots when they were not told which were AI) and detection (recruiters spotted the AI headshots 39.5% of the time), not likeness. Magic Studio, which sells an AI headshot service, wrote in a June 2025 comparison of ten services that scoring likeness means feeding every tool the same photos and having people who do not know which tool made which image rate the results, and that "Nobody has published that test." Connor Kilrain and colleagues (December 2025) proposed ranking generated images against a gallery of real ones to measure lost identity, but tested it on birds, cars and animals, not on people's headshots. A search of arXiv for headshot studies found none of commercial services.

ringover.commagicstudio.comarxiv.org

Built with what exists

Trial packs from commercial AI headshot services; InsightFace's open ArcFace models for face matching; the gallery-ranking method of Kilrain and colleagues; a web form for blind lineups.

  1. First step

    Put five consenting volunteers through one service and build the real-photo gallery from all five.

  2. Done looks like

    30 volunteers, four services, and for each service the share of headshots that the model ranks as the right person first and the share that acquaintances pick correctly, set beside the same two scores for each volunteer's real photograph.