Why we built the tool in the first place

The trigger was mundane: we wanted the compressor to show which camera had taken a photo. One look at the EXIF data of real test photos quickly turned that into something else — the realization of how much a smartphone photo actually reveals about its owner. Out of that grew the image info tool: show everything an image file contains, and make removing it as easy as viewing it.

Finding 1: GPS is more precise than you think

We knew photos contain location data. What surprised us: the precision. Modern smartphones don't just write latitude and longitude with several decimal places (that's meter-level accuracy — enough to pinpoint your front door), but often also the altitude above sea level and the direction the camera was facing. From a single carelessly shared photo, you can reconstruct which floor it was taken on. Ever since, our tool displays GPS findings with a clear warning right at the top — and the same warning made its way into our guides, from the WhatsApp post to the classified-ads rules.

Finding 2: the orientation flag — the secret final boss

The most instructive mishap of the development: test images from an iPhone reliably appeared in the tool rotated by 90 degrees. The reason is a piece of photography history: cameras often store the pixels the way the sensor delivers them — sideways — and merely note in an EXIF field called Orientation which way up the image is meant to be. Every program that displays images has to read this field and apply the rotation itself. Modern browsers now do this automatically; but if you draw pixels directly onto a canvas, under certain circumstances you get the raw view.

For us that meant: build in explicit orientation handling and test with photos taken in all four holding positions. For you it means: if an image appears correct in one program and rotated in another, that one byte is almost always to blame — and re-exporting through any editing tool (which "bakes in" the rotation) fixes the problem for good.

300 × 250 — Rectangle
Cookie-Banner ausstehend

Finding 3: serial numbers and software traces

Less well known than GPS: many cameras write their device serial number into the EXIF data, some even the lens's. That makes it possible to attribute photos to the same device — across years and platforms. On top of that come software traces: editing programs leave behind names and version numbers. An "unedited" photo whose metadata names Photoshop has settled more than one argument.

Finding 4: PNG plays by different rules

One expectation we had to correct: PNG has no EXIF in the classic sense. Screenshots — the most common PNG type — are therefore metadata-poor by default. But not metadata-free: PNG has its own text blocks, and macOS screenshots, for instance, record software information there. The rule of thumb we took away from it: photos = lots inside, screenshots = little inside, but "little" is not "nothing".

How the tool works today

For reading we use the open-source library exifr, which parses EXIF, GPS, and camera blocks directly in the browser; for writing and removing, piexifjs does the job. Both run — like everything we build — locally: for a tool that displays sensitive location data of all things, a server upload would be absurd. The background of that architecture decision is covered in the workshop post Why your images never see our servers.

300 × 250 — Rectangle
Cookie-Banner ausstehend

What we've been telling users ever since

  • Before sharing with strangers: run it once through the metadata tool — looking takes ten seconds, and removing is usually the right call.
  • Don't trust platforms blindly: the big social networks strip EXIF on upload; email and cloud folders don't.
  • Metadata isn't evil: capture date and camera settings are valuable for your own photo collection. The question is never "metadata yes or no" but "who gets to see it".

Sources

CIPA — EXIF standard (official specification) · exifr — source code on GitHub · MDN — image-orientation.