From Postcards to Pixels: A History of Travel Photography
How the way we remember a trip has changed from a postcard bought at a kiosk to an AI guessing where our own photo was taken.
Short answer
The history of travel photography runs from bought postcards, through rationed rolls of film, to camera rolls holding thousands of unsorted images. Each step made pictures cheaper and more personal while leaving more of them unlabelled, which is why guessing where an old holiday photograph was taken is now a hobby.
For most of the history of travel, the photograph you brought home was not yours. It was a postcard, printed by the thousand from someone else's negative, chosen off a spinning rack because it looked the way your trip was supposed to look. Somewhere along the way that changed completely: travel photography became personal, then constant, and now, increasingly, something a model helps us make sense of after the fact.
It is a strange arc when you lay it out end to end. Buying a professional's photograph of a place, then taking your own, then taking thousands of your own, then needing help working out which thousand was which. Here is how the souvenir photograph got from a kiosk rack to a camera roll nobody can fully label.
What did travellers bring home before cameras were portable?
A bought image. The picture postcard was the dominant souvenir: a local studio photographed the landmark once, printed it at scale, and travellers posted a copy home. The trip was represented by a photograph the traveller had no hand in making.
Before cameras were ordinary travel kit, the postcard was how most people documented — or really, purchased evidence of — having been somewhere. A photographer, usually a local studio, shot the landmark once, printed it at scale, and travellers bought a copy to post home or paste into an album. The image was standardised, idealised and almost never contained the traveller. The picture postcard boom peaked in the years around 1905 to 1915, when the format was cheap enough and the post fast enough that a card could reach home before the sender did.
This made the visual record of travel oddly impersonal for decades. Near-identical postcards of the same tower or coastline circulated in enormous numbers while the actual lived experience of a trip — the wrong turn, the strange cafe, the person you met on a train — went entirely undocumented, because documenting it yourself required equipment most people did not carry.
How did film change what people photographed?
It put the camera in the traveller's hands but kept photographs scarce. A roll held 24 or 36 exposures and both film and developing cost money, so people composed carefully, took one attempt at a scene, and saw the results weeks later.
Portable film cameras changed the equation slowly and with real friction. Every frame cost money to buy and money to develop, so people rationed shots with a discipline that now seems unthinkable. You did not photograph your lunch. You photographed the landmark, perhaps your companions in front of it, and saved the rest of the roll for whatever came next. Disposable cameras in the 1980s and 1990s lowered the barrier again, letting casual travellers document a trip without owning equipment, but the underlying scarcity held: a finite number of frames, developed long afterwards, usually far from the place they showed.
That era produced something whose effects are still with us — boxes of undated, unlabelled prints in attics and drawers, taken by someone who reasonably assumed they would always remember where each one was. They usually did not, and the job of working it out afterwards is pure visual detection, the same accumulation of small clues traced in visual deduction from Sherlock Holmes to AI.
Digital removes the cost of curiosity
Digital cameras, and then camera phones from 2000 onwards, broke the scarcity model entirely. Once a photograph cost nothing to take and nothing to develop, the rationing instinct disappeared. People began photographing everything: not just the landmark but the street it stood on, the meal beside it, the sign nobody could translate. Phones in particular made photography ambient rather than occasional — something happening constantly in the background of a trip rather than a deliberate act reserved for photogenic moments.
Phones also appeared to solve the labelling problem the film era never could, embedding coordinates directly into a photograph's metadata at the moment of capture. For the first time an image could carry its own location without anyone doing anything. The fix turned out to be more fragile than it looked.
Why do so many holiday photos still have no location?
Because the coordinates sit in a metadata block that is easily lost. Messaging apps and social platforms strip it on upload, plenty of people switch location tagging off, screenshots never carry it, and every scanned print from the film era never had it at all.
The location lives in the file's Exif block, and that block is the first thing to go when an image travels. Send a photograph through a messaging app and the copy that arrives is usually clean. Save a screenshot of it and the coordinates were never there in the first place. Export it through the wrong editing tool and the block quietly disappears. The pile of geographically orphaned photographs did not shrink when phones learned to geotag. It moved from shoeboxes to camera rolls and old hard drives, and it kept growing, because far more pictures are taken now than ever were printed.
From taking photos to making sense of them
That is what the current chapter is really about. Nobody is short of travel photographs. Most people have tens of thousands across a camera roll, a cloud backup and an inherited family archive, many with no caption and no metadata to lean on. The interesting problem has flipped from how do I capture this trip to how do I make sense of everything I already captured.
This is where AI-assisted photo geolocation arrives, as a genuinely new tool in a very old story. Raven lets you upload a picture — a scanned print from a parent's album, an unlabelled holiday shot, anything with no location attached — and Google's Gemini vision model reads the architecture, vegetation, signage and light in the frame to offer its best guess about where it might have been taken. Nothing is kept: the image is processed in memory for that one guess and then discarded. The longer version of how that capability arrived is in a short history of photo geolocation, and the parallel shift in how we shoot is covered in how technology changed the travel photo.
What can a model actually tell from an old holiday photo?
As much as the frame contains. Legible signage, a distinctive roofline, period vehicles or a named shopfront can support a specific guess. A beach, a plain wall or a hotel interior often supports nothing narrower than a continent, and an honest tool will say so.
The useful mental model is that the photograph sets the ceiling, not the software. Shots that were meant to be beautiful are frequently the hardest, because a photographer framing a sunset deliberately excludes the road signs, shopfronts and utility poles that carry the geographic information. The accidental photographs — the one of your cousin taken in front of a bus stop — are often far more informative. When the evidence genuinely points several ways at once, the right answer is a region rather than a pin, which is the case examined in how AI handles photos with multiple possible locations.
Dig out one unlabelled print, scan it, and see what the frame still gives away.
Upload a photo →It is worth sitting with the shape of that arc: a photograph you did not take, then a photograph you rationed, then a photograph that cost nothing at all, and now a photograph you might need a little help remembering the story behind. The postcard rack is long gone, but the question printed on the back of every one of those cards — where exactly was this — has followed the medium through every format since.
Frequently asked questions
- Why did people buy postcards instead of taking photographs?
- Cameras were expensive, heavy and slow for most of the nineteenth and early twentieth centuries. A printed postcard of the landmark was cheaper, sharper and immediately available at the kiosk beside it.
- How many photographs did a film traveller take on a trip?
- Usually a few dozen. A roll held 24 or 36 exposures and both the film and the developing cost money, so most people carried two or three rolls for a fortnight and rationed them carefully.
- Why are so many old holiday photos unlabelled?
- Because labelling was optional and the photographer assumed they would remember. Prints were developed weeks after the trip, often far from the place, and captioning the back was a habit only some families kept.
- Can an AI work out where an old travel photograph was taken?
- It can offer a guess from what is visible: architecture, signage, vehicles, plants and light. Rich street scenes often produce a plausible region; a faded portrait against a plain wall produces very little.
Sources
- Postcard — WikipediaThe picture postcard boom peaked in the years around 1905 to 1915, when billions were posted worldwide.
- Camera phone — WikipediaThe first commercial camera phones reached consumers in 2000, the point at which travel photography became ambient rather than planned.
- Exif — WikipediaThe metadata block that carries GPS coordinates in digital images, and the one platforms routinely strip on upload.
Reminder
Raven is built for entertainment and curiosity. Its guesses are AI estimates that can be wrong, and it must never be used to track or identify real people. Uploaded photos are processed in memory and immediately discarded — never stored.


