If every camera on Earth vanished, the world would lose a machine that takes about 2.1 trillion photos a year, and memory, news, science and courts would fall back on what they used before 1839: drawing, writing and eyewitness description, which research shows are slower and far less reliable. But the loss could not last. A camera is only a dark box with a hole and a light-sensitive surface, and the straight-line travel of light means any darkened room with a pinhole starts projecting an image again the same afternoon.
- 2.1 trillion photos were taken in 2025 by Photutorial's estimate, about 94% of them on phones.
- 4.4 billion smartphone image sensors were shipped in 2024 alone (Counterpoint), before counting cars, CCTV and medicine.
- About 11 minutes a day: what each person on Earth would spend sketching to replace that output at 15 minutes a drawing (my estimate).
- 69% of US DNA exonerations involved an eyewitness who picked the wrong person, which is what proof looks like without pictures.
- 0.47 mm: the ideal pinhole for a 100 mm box camera, by Rayleigh's formula. That is all it takes to rebuild one.
The question sounds like a writing prompt, but it is a useful test. It shows how much of modern life rests on a cheap piece of silicon, and which older human skills would be called back to work. I'll take it in four parts: the scale of what disappears, what replaces it, what the replacements cost, and how fast physics hands the camera back.
How many cameras would actually disappear?
Far more than the phones in our pockets. Counterpoint Research counted about 4.4 billion image sensors shipped in smartphones in 2024, an average of 3.7 cameras per phone, and that excludes car cameras, CCTV, webcams, satellites, factory inspection, endoscopes and the dashboard cameras of every delivery van.
The output is just as large. Photutorial estimates about 2.1 trillion photos were taken in 2025, roughly 5.7 billion a day, up from 660 billion in 2013. Phones take about 94% of them; dedicated cameras, at 9.44 million shipped in 2025 by CIPA's count, are a rounding error.
See the data as a table
| Photos taken ( trillion) | |
|---|---|
| 2013 | 0.66 |
| 2014 | 0.88 |
| 2015 | 1.0 |
| 2016 | 1.1 |
| 2017 | 1.2 |
| 2018 | 1.33 |
| 2019 | 1.44 |
| 2020 | 1.15 |
| 2021 | 1.2 |
| 2022 | 1.71 |
| 2023 | 1.81 |
| 2024 | 1.94 |
| 2025 | 2.1 |
Source: Photutorial, Photo statistics (updated September 2026) · Estimates built on Rise Above Research forecasts, CIPA and IDC shipments. 2020 dips with lockdowns.
What stops working on day one
- Memory: the family album, the phone gallery and the photo of where you parked.
- News: every live broadcast, press photo and video clip.
- Proof: CCTV, body cameras, dashcams, passport photos and face unlock.
- Science and medicine: telescopes with sensors, microscope cameras, endoscopy, satellite weather imagery.
- Industry: barcode scanners, parking systems, robot vision and car driver-assist cameras.
What did the world use before cameras?
Hands and words. For most of history a likeness meant a painter, an event meant a written account, and a criminal meant a witness's description. Those skills still exist in pockets, and they show what a camera-free world would look like.
Until 1862, Americans saw their war mostly through artists' impressions. On 19 September 1862, two days after the Battle of Antietam in Maryland, the bloodiest day of combat on US soil with more than 23,000 killed, wounded or missing, Mathew Brady sent the photographer Alexander Gardner and his assistant James Gibson to the battlefield. In October 1862 Brady hung Gardner's pictures in his New York gallery under the title The Dead of Antietam, and The New York Times reviewed the show on 20 October. Many of the images were graphic photographs of corpses, something new to America. For many people it was the first time they saw the reality of war in photographs rather than in an artist's version of it. Take every camera away and the record goes back to that version.
Wikipedia: Mathew Brady (The Dead of Antietam, 1862)
Courts already run a camera-free experiment
Many courtrooms ban cameras, so news organisations hire sketch artists. In the UK and Hong Kong, artists may not even draw during proceedings: they watch, leave and draw from memory or notes. In the US, Rule 53 of the Federal Rules of Criminal Procedure has kept broadcast out of federal criminal courts since the Hauptmann trial era. The New York Times was printing courtroom sketches as early as 1889.
This tells us the replacement exists and works, slowly. A professional courtroom sketch can be done in about eight minutes, and it carries a likeness, a mood and an arrangement, but not a measurement anyone could use as evidence.

How much drawing would it take to replace photography?
About ten minutes per person per day, at the most generous assumptions. Here is the working, so you can change any number you disagree with.
- Define the output. Photos per year \( P = 2.1\times10^{12} \) (Photutorial, 2025).
- Define the workforce. Everyone. World population \( N = 8\times10^{9} \), the UN milestone of November 2022.
- Per head. \( P/N = 2.1\times10^{12} / 8\times10^{9} \approx 262 \) pictures per person per year.
- Time per picture. Assume a quick sketch takes \( t = 15 \) minutes, my estimate between an 8-minute courtroom sketch and a careful study.
- Daily load. \( 262 \times 15 / 365 \approx 10.8 \) minutes per person per day.
That looks manageable until you remember who would do it. Most people cannot draw a recognisable face in fifteen minutes, so the real load falls on a small trained group. If one person in a thousand is a capable illustrator, each would face about 180 hours a day of work, which is impossible.
Sensitivity check
- At the courtroom pace of 8 minutes, the load falls to about 5.8 minutes per person per day.
Calculated in this post from Photutorial estimate (about 2.1 trillion photos in 2025) and the UN figure (8 billion people), with these example inputs: Wikipedia, Courtroom sketch, the pace of one sketch drawn in about eight minutes. 2.1 × 1012 × 8 / (8 × 109 × 365) ≈ 5.8 minutes.
- At a one-hour portrait, it rises to about 43 minutes per person per day.
Calculated in this post from Photutorial estimate (about 2.1 trillion photos in 2025) and the UN figure (8 billion people), with these example inputs: a one-hour portrait, an assumed pace for this check and not a measured figure. 2.1 × 1012 × 60 / (8 × 109 × 365) ≈ 43 minutes.
- If only 1% of photos were worth replacing (most phone photos are near-duplicates), the 15-minute case falls to about 6.5 seconds per person per day.
Calculated in this post from Photutorial estimate (about 2.1 trillion photos in 2025) and the UN figure (8 billion people), with these example inputs: 15 minutes a picture and 1% of photos replaced, both assumptions of this post. 10.8 minutes × 0.01 ≈ 0.108 minutes, about 6.5 seconds.
The last line is the real answer. A camera-free world would not draw two trillion pictures; it would draw a few billion and simply stop recording the rest. What vanishes is not the art, it is the casual record.
| Job a camera does | Pre-1839 replacement | What gets worse |
|---|---|---|
| Family memories | Diaries, letters, painted portraits for the rich | Most people leave no likeness at all |
| News | Reporters' text, engravings, sketch artists | Delay of days; images can't be checked |
| Criminal evidence | Eyewitnesses, descriptions, fingerprints | Misidentification becomes the main risk |
| Identity documents | Written descriptions of height, scars, eye colour | Forgery is easy |
| Science | Drawings at the eyepiece, measurements by hand | Fast and faint events go unrecorded |
| Art | Painting from life and memory | Errors like the flying gallop return |
Would we remember more without cameras?
Possibly, a little, for the moments themselves. Psychologists have a name for the opposite effect: the photo-taking impairment effect, where people remember photographed objects worse than ones they simply looked at, as if the camera took the job of remembering off their hands.
Linda Henkel first showed it on a museum tour in 2014: students recognised photographed objects less accurately and answered fewer detail questions about them. Zooming in on a detail removed the effect, which points to attention as the mechanism. Later studies repeated it in labs, and the most rigorous test so far is below.
Three experiments with 60, 96 and 243 university students had participants view paintings and either only observe them, photograph each once, or photograph each five times, then answer four-option visual detail questions (chance is 25%). In the between-subjects Experiment 2, both photo groups did worse than observers, with large effects: d = 0.72 for one photo and d = 1.27 for five. Telling people to take five unique photos did not rescue memory. The limit: these are short lab tests on students looking at paintings on screens, not long-term memory of real life events, and photos you later review can help memory rather than hurt it.
Read the paper (free)So the vanishing would trade two things. In the moment, people might attend more closely. Over a lifetime, they would lose the external record, because human memory reconstructs rather than replays. A photo is the reason many of us can picture a grandparent's face at thirty.
What would replace photographs as proof?
Description, and that is the weak point. The Innocence Project reports that 69% of DNA exonerations in the United States, 252 of 367 cases, involved eyewitness misidentification. Those were convictions in a world that had cameras; without them, a witness's account would carry even more weight.
Research on how to make descriptions better is active, because police still rely on them. Describing a face in words is now known to help, not hurt, memory in most conditions.
209 undergraduates studied two blocks of 12 faces and, after each block, either described the faces (by parts, such as nose and eyes, or by traits, such as "friendly") or did a counting control task. Both kinds of description improved later recognition compared with the control, an effect called verbal facilitation, which runs against the older "verbal overshadowing" worry. Trait descriptors raised accuracy without inflating confidence, while featural descriptions changed how well confidence tracked accuracy. The limit: faces were studied in a lab, not seen under stress at a crime scene.
Read the paper (free)A photograph is a witness that does not get tired, scared or persuaded. Take it away and every other witness becomes more important.From this post
The irony of 2025 and 2026
Photographs are already losing their power as proof, cameras or not, because AI images look real. Camera makers now sign pictures at the moment of capture using the C2PA standard so a file's history can be checked. Legal scholars warn of the "liar's dividend": once fakes are common, real footage can be dismissed as fake. A camera-free world would be an extreme version of a problem we already have.
How fast could we rebuild a camera?
Within a day for the image, within days for a permanent picture. The camera obscura needs no invention, only the fact that light travels in straight lines. Mozi described the effect in China around 500 BCE, and Ibn al-Haytham studied it in the eleventh century.
Worked example: the best pinhole and the exposure
Lord Rayleigh's 1891 rule balances two blurs: a big hole lets each point spread into a disc, a tiny one spreads it by diffraction. The optimum diameter grows with the square root of focal distance times wavelength.
Now the exposure. The sunny 16 rule says that in bright sun at f/16 the right shutter time is one over the ISO. Each doubling of f-number cuts light by four, so time scales with the square of the ratio.
With ISO 100 film, a shoebox with a needle hole takes a sharp-enough picture of a sunlit street in under two seconds. With a slow, home-made material the time stretches to minutes, which is exactly what Louis Daguerre faced. His Boulevard du Temple of 1838 needed about four to five minutes, so the busy Paris street looks empty except for one man standing still to have his boots polished.

Use this to check what you just read: set a box depth and a film speed and see the pinhole size and the exposure in full sun.
Rebuild a camera: pinhole size and sunny exposure
A camera-free world is livable: courts, newspapers and families ran on sketches, text and witnesses before 1839, at the cost of slower news, weaker evidence (eyewitness error already appears in 69% of US DNA exonerations) and no casual record of ordinary lives. Replacing even a fraction of today's 2.1 trillion annual photos by hand would take every person about 11 minutes a day. But the absence cannot hold, because the rectilinear propagation of light makes any dark room with a 0.5 mm hole a camera, and with ISO 100 film it takes a sunlit picture in about two seconds.
Would painting get its old job back?
Yes, and that is the most interesting consequence for artists. Photography freed painting from the duty of recording likeness, which is part of why painters after the 1840s moved towards impression, abstraction and expression. Remove the camera and portrait painters, illustrators and engravers would be essential workers again.
- Illustrated news would return, like the engraved illustrated papers of the Victorian era, made from artists' field sketches.
- Printmaking would boom, because one drawing would need thousands of copies. That is exactly the business Raja Ravi Varma built in the 1890s, as told in the story of his press.
- Seeing skills would be taught again. Composition advice like the rule of thirds came from painting, and the gap between what the eye sees and what a lens records, covered in where the eye and camera lie, would matter to everyone.
Photojournalism would lose its documentary power first. Pioneers like Homai Vyarawalla, India's first woman press photographer, built a visual record of the freedom years that no sketch could match for trust.
What's new
- Sony extends its C2PA camera authenticity signatures from stills to video for news organisations ProVideo Coalition, Oct 2025
- What is truth in a post-photography world: C2PA gaps and the liar's dividend Fstoppers, May 2026
- Photo count update: about 2.2 trillion photos expected in 2026 Photutorial, Sep 2026
- Smartphone image sensor shipments reach 4.4 billion, with fewer cameras per phone Counterpoint Research, Apr 2025
What if next?
- What if you had to design a logo that lasts 10,000 years? Without cameras or common language, how do you warn the future about buried nuclear waste?
- What if Raja Ravi Varma had a camera and Photoshop? The reverse experiment: give a nineteenth-century painter every tool we have.
- What if you took photos on Mars? Cameras that work up to 400 million km away, and why their colours need a calibration target.
Questions people ask
What did people do before cameras existed?
They hired painters for portraits, used engravings and written accounts for news, and relied on witnesses for evidence. Most ordinary people left no picture of themselves at all.
Can you make a camera with no lens?
Yes. A pinhole camera uses a tiny hole instead of a lens. For a 100 mm deep box the ideal hole is about 0.47 mm wide, giving roughly f/213; in sunshine with ISO 100 film the exposure is about two seconds.
Does taking photos make you remember less?
In lab and museum studies, yes for the photographed items: people recalled fewer visual details. Paying deliberate attention, such as zooming in on a detail, reduced the effect.
Why are cameras banned in some courtrooms?
To avoid distraction and protect privacy. In the UK and Hong Kong, artists may not sketch during proceedings either, so they draw from memory afterwards.
The Kala world page has more on design, photography and how to see.
Sources
- Photutorial: Photo statistics (2026) (photos per year 2013–2026, 94% smartphone share, CIPA shipments)
- Counterpoint Research: Global smartphone CIS shipments climb 2% in 2024 (2025) (4.4 billion sensors, 3.7 cameras per phone)
- IEEE Spectrum: June 1878, Muybridge photographs a galloping horse (date, Sallie Gardner, 1/1,000 s shutter, Géricault comparison)
- Wikipedia: Mathew Brady (story: Gardner and Gibson sent 19 Sep 1862, 23,000+ casualties, The Dead of Antietam exhibition Oct 1862, NYT review 20 Oct 1862)
- Soares and Storm, Psychonomic Bulletin & Review: Does taking multiple photos lead to a photo-taking-impairment effect? (2022) (sample sizes, effect sizes, Henkel 2014 summary)
- Association for Psychological Science: No pictures, please (Henkel 2014) (museum tour design, zoom finding)
- Weatherford et al., Scientific Reports: Different types of facial description alter the confidence-accuracy relationship (2026) (verbal facilitation, N = 209)
- Innocence Project: How eyewitness misidentification can send innocent people to prison (69%, 252 of 367 DNA exonerations)
- Wikipedia: Courtroom sketch (UK and Hong Kong rules, Rule 53, 1889 sketch)
- Wikipedia: Pinhole camera (Rayleigh 1891 formula, camera obscura history)
- Wikipedia: Sunny 16 rule (exposure rule used in the calculation)
- Wikipedia: Boulevard du Temple (photograph) (1838, four-to-five-minute exposure)
- United Nations: Population (8 billion milestone, November 2022)
- ProVideo Coalition: Sony launches first video-compatible camera authenticity solution (2025)
- Fstoppers: What is truth in a post-photography world? (2026) (C2PA adoption, liar's dividend)
