The naive way (and why it's not ideal)

The obvious approach is to draw the image onto a <canvas> and export it again. Canvas output has no metadata — but it re-encodes the JPEG, so you lose a little quality every time.

The better way: drop the metadata segments

A JPEG is just a sequence of marker segments. Each starts with 0xFF then a marker byte. EXIF/GPS live in the APP1 segment (0xFFE1); other APPn and comment (COM) segments hold similar metadata. The actual image data starts at the SOS marker (0xFFDA).

So the algorithm is simple: walk the segments from offset 2, and for each marker read its 2-byte length. If the marker is APP1–APP15 (0xE1–0xEF) or COM (0xFE), skip it; otherwise keep the segment as-is. When you reach SOS (0xDA), copy the entire rest of the file and stop. Concatenate the kept parts and you have a valid JPEG with the metadata gone — byte-for-byte identical image data, no quality loss.

That's about 15 lines of plain JavaScript with Uint8Array.subarray — no library required. To show users what's being removed, read the tags first with a small library like exifr (GPS, make/model, date), then strip.

Why do this client-side?

The point is privacy — you shouldn't have to upload a photo to a server just to remove its location. Everything runs in the browser, so the image never leaves the device. I turned this into a free image metadata remover.

Anyone else strip metadata before posting? Curious what edge cases you've hit.