EXIF Remover

JPEG, PNG and WebP all keep metadata in discrete container structures — APPn segments, PNG chunks, RIFF chunks — that sit beside the compressed pixel data rather than inside it. Removing metadata therefore means rewriting the container's table of contents and copying the pixel bytes across verbatim. Nothing is decoded, nothing is re-compressed, and the picture is byte-for-byte the same image it was before.

The common shortcut is to draw the photo onto an HTML canvas and export it again. That does remove metadata, because the canvas knows nothing about it, but it also decodes and re-compresses the whole image, losing a little quality every time, and it silently discards transparency and animation along the way. It is the reason a supposedly simple metadata removal can come back looking softer than the original.

A photo straight off a phone commonly records the exact coordinates where it was taken, accurate to a few metres, the precise date and time, and the device make and model. Cameras add body and lens serial numbers, which link every photo taken with that equipment together even when nothing else identifies them. Editing software adds its own layer: XMP can carry a creator name and an edit history, and IPTC fields hold captions, credits and copyright.

None of that is visible when you look at the picture, which is what makes it easy to overlook. A photo posted to a forum, attached to a marketplace listing or sent to someone you have not met can carry your home address without any indication on screen.

Phones do not usually rotate the pixels when you turn the handset. They save the sensor's native landscape frame and add an orientation tag telling the viewer to rotate it for display. That tag lives in EXIF, so a remover that deletes all EXIF also deletes the instruction, and every portrait photo suddenly displays on its side.

This is handled by writing back a minimal EXIF block containing the orientation value and nothing else — 26 bytes, no camera, no timestamp, no location. The photo displays correctly and still carries nothing that identifies you. If the image is already upright the block is skipped entirely rather than added as dead weight, and you can switch the behaviour off if you want every last tag gone.

It is easy for a tool to claim a file is clean. The result panel here is generated by parsing the finished file again and reporting what is actually in it, so the summary reflects the download rather than an assumption that the operation worked. For an independent check, put the cleaned file through the EXIF Viewer — it should report no metadata at all, or the single orientation tag if you chose to keep it.