How to find anything in your camera roll
Your phone can already read the words inside your pictures, and most people never use it. Here's how — and where it stops working.
The short version: open Photos, use the search box, and type a word that was printed in the
picture — wifi, a brand, a street name. Both iPhone and Android read text inside images and have
done for years. It works far better than most people expect, and it fails in one specific way that’s
worth understanding.
On iPhone
Photos search covers three different things at once, and knowing which is which saves a lot of guessing.
Text in the picture. Type password, invoice, a brand name, a street name. This is Live Text,
and it reads printed words in your photos. If you photographed a router sticker, searching wifi
usually finds it.
Things it recognises. dog, car, receipt, menu, whiteboard. Not read from text —
recognised from the image itself.
Where and when. London, June, a person’s name once you’ve tagged them.
You can stack them. receipt London narrows to both. And on a single photo, long-press any text in
it to copy — that’s how you get a serial number out without typing it.
On Android
Google Photos does the same thing, and has since 2019. Search a word that appears in the image and it comes back. Its launch pitch for the feature named exactly the cases people use it for: wifi passwords saved as photos, business cards, receipts.
Google Lens goes further on a single image — point it at a photo and it’ll pull out text, translate it, or look up a product. That’s for the picture in front of you, not for searching your library.
Where it stops
This is the part that isn’t in the tips articles.
You have to remember a word that was in the frame. Search is only as good as your memory of the
printing. “The green paint from the hallway” finds nothing, because the tin didn’t say hallway and
it probably didn’t say green either — it said Farrow & Ball Breakfast Room Green, and if you
could remember that you wouldn’t be searching.
It gives you the picture, not the answer. You still have to open it, find the number, read it correctly, and type it somewhere. On a plate that’s been behind a cupboard for six years, in bad light, at an angle, that’s not a small step.
It can’t do collections. There’s no way to ask for every paint you’ve photographed, or every appliance whose warranty is running out. Those are questions about a set, and a photo library is not a set — it’s a timeline.
Handwriting and awkward angles beat it regularly. Curved surfaces, glare, and anything written by hand are hit and miss.
Practical things that help right now
- Favourite it at the moment you shoot. One tap, and the useful ones land in an album you can actually scroll.
- Use the Screenshots album as a starting point rather than the whole roll — it’s already a separate smart album on both platforms.
- Search the shop, not the thing. You’ll more often remember the retailer’s name than the product code, and the receipt has both.
- Long-press to copy text rather than retyping a serial. Fewer mistakes.
When the search stops being enough
The pattern is consistent: photo search is good at finding a picture and bad at giving you a fact. That’s not a flaw, it’s what it was built for.
Keeplea works the other way round. It reads the thing when you capture it and keeps the details as text — the model, the serial, the colour code, the network name — so asking which paint was in the hallway returns the paint, with its numbers, rather than a grid of thumbnails to squint at. And because those captures never enter the camera roll, the roll stays the thing you scroll for pleasure.
Try the search first. Genuinely — most people have never used it and it solves more than they think. If you find yourself doing it often and coming back with the right photo and the wrong outcome, that’s the gap worth fixing.