ResizeTo

Guide

What EXIF metadata reveals

Every JPEG carries a block of metadata in front of the picture: camera model, lens, exposure settings, the software that edited it, several timestamps, and on most phone photos the coordinates. One camera original taken apart here held 41 such fields.

What is actually in there

JPEG files are a chain of labelled segments. The picture is one of them; the rest carry metadata. Taking apart one camera original, 8.9 MB from a Canon DSLR edited in Lightroom, gives this:

SegmentBytesWhat it holds
APP1 / Exif12 460Camera, lens, exposure, timestamps, orientation
APP1 / XMP9 051Adobe's edit history and cataloguing fields
APP13 / IPTC11 698Captions, keywords, copyright, creator
APP2 / ICC3 162Colour profile
APP14 / Adobe16Colour transform marker

36 387 bytes of metadata, 0.41% of the file.

The EXIF block alone held 41 fields across two directories. Among them, in plain readable text: the manufacturer, the exact camera model, and a software string naming the editing application, its version number, and the operating system it ran on. Plus the shutter time, the aperture, the ISO, the lens, and the date and time to the second.

None of that is sinister on its own. Together it is a fingerprint. Two photographs carrying the same camera model, the same lens and the same software build were, with high probability, taken by the same person.

Many cameras go further and write a body serial number and a lens serial number. Those are unique. If one photograph of yours is public and attributable, every other photograph from that body can be linked to it.

The coordinates

The GPS fields sit in their own directory inside the EXIF block. Latitude and longitude are each stored as three rational numbers for degrees, minutes and seconds, with a separate field saying north or south, east or west. Phones typically write them to a precision of a few metres.

Alongside them you will often find the altitude, a GPS timestamp independent of the camera clock, and the compass direction the camera was pointing.

A photograph of a living room, posted publicly, can carry the street address of that living room to within the width of the building. This is the ordinary default behaviour of a phone with location services enabled for the camera.

What a real folder looks like

44 JPEGs over 40 KB were scanned from an ordinary laptop: downloads, screenshots, photos received from other people.

The zero is the interesting number, and it is easy to misread. It does not mean phones are not writing coordinates. It means that every one of those files had passed through something that removed them on the way: a social network, a messaging app, a website's upload pipeline.

So your Downloads folder cannot answer the question. What decides the outcome is the route a given file is about to take, because that route is what does or does not strip the coordinates.

Who strips and who does not

The pattern is consistent, if unwritten. Anything that re-encodes your image for its own purposes drops the metadata as a side effect. Anything that treats the file as a file passes it through untouched.

That last one deserves emphasis, because the instinct is backwards: choosing "send as file" to avoid losing quality is exactly the choice that preserves the coordinates too.

The thumbnail that did not get the memo

EXIF can embed a small JPEG preview of the image, written when the photo was taken. Some editing software updates the picture and forgets the preview. The result is a file whose visible image has been cropped or retouched and whose hidden thumbnail still shows the original.

It is an old bug, it has produced real disclosures, and it is a good argument for removing metadata rather than editing it selectively.

Removing it is not a size optimisation

At 0.24% of file weight in that sample, and 0.41% even on a metadata-heavy camera original, stripping EXIF will not get you under an upload limit. Anyone offering it as a compression feature is selling you a rounding error.

Remove it for what it is: a privacy operation, and a small one to perform.

The one field you should not lose carelessly

EXIF carries an orientation flag. Phones do not rotate the pixels when you turn the handset. They record the picture as the sensor saw it and note which way is up. Every viewer is then supposed to read the flag and rotate on display.

Strip the metadata without acting on that flag and the image is suddenly sideways. This is the single most common way a "clean my photo" step goes wrong, and it is why some tools produce rotated files.

The right behaviour is to apply the rotation to the actual pixels before dropping the tag, so the image is upright by construction and needs no flag at all. That is what this site does: the decoder is asked for the picture already oriented, and everything downstream works on upright pixels.

Getting rid of it

Any genuine re-encode removes metadata, because the encoder writes a new file from pixels alone and has nothing to copy across. Compressing a photo here, or resizing it, leaves nothing behind but the ICC-less pixels: no camera, no serial number, no coordinates. The metadata row on the tool pages is set to Remove by default, so that re-encode happens even when the file already meets your limits.

If you would rather look before you leap, the metadata viewer reads the file in your browser and shows you what is in it, including a warning when coordinates are present. Nothing is uploaded, which matters more than usual here: sending a file somewhere to find out whether it discloses your location is a strange way to protect it.

Strip metadata by compressing? Compress to a file size →

Related guides