What EXIF is
EXIF (Exchangeable Image File Format) is a structured block of metadata stored alongside the pixels in a JPEG, HEIC or TIFF file, and optionally in PNG. It is written by the camera or phone at capture time and can be edited later. It covers the camera make and model, the lens, the exposure triangle (aperture, shutter, ISO), the focal length, the capture timestamp, an orientation flag, and — when location services are on — a GPS block with latitude and longitude.
The metadata is not encrypted or hidden. Anyone with the file can read it with the right tool, and many operating systems show some of it in the file's properties dialog. That is the whole privacy issue in one sentence: a photo you share may tell the recipient where it was taken even if the image itself contains nothing identifiable.
What a browser can read, measured
The inspector on this site reads the common EXIF tags from a chosen file without uploading it. To make the scope concrete, here is a 640 × 480 JPEG written with five EXIF fields — Make, Model, Orientation, DateTime and a GPS block — and the tags the inspector reports for it:
| Field written to the file | Value | Shown by the inspector |
|---|---|---|
| Make | TestCam | Yes — TestCam |
| Model | Model X | Yes — Model X |
| Orientation | 6 | Yes — Rotated 90° CW |
| DateTime | 2026:10:24 12:00:00 | Yes — Date/time |
| GPSInfo | Latitude/longitude block | No — the coordinate block is deliberately not decoded |
The point of showing the last row is honesty about scope. The viewer lists the common tags; it does not turn the GPS block into a map. If your question is “does this photo contain my location”, assume a phone photo does and remove the metadata before sharing rather than relying on the tag list.
Orientation: the tag that changes what you see
Orientation is the most practically confusing EXIF field. A phone often stores the sensor's native landscape pixels and records how the device was held, so the viewer can rotate the image for display. The tag values are not angles: 1 is normal, 3 is 180°, 6 is 90° clockwise, 8 is 90° counter-clockwise, and 2, 4, 5, 7 add a mirror. A photo that looks sideways in one program and upright in another is almost always a file whose orientation tag is being honoured by one and ignored by the other.
This matters the moment you process an image. Resizing or re-encoding through a canvas draws the
decoded pixels, and browsers apply the orientation flag when decoding an <img>, so
the output is usually upright — but the tag itself is gone. If a tool ignores the flag, the same file
comes out sideways. Check the result, not just the input.
What survives a re-encode
Canvas-based processing — the technique every image tool on this site uses — draws the decoded pixels
into a new bitmap and asks the browser to encode that bitmap. No metadata is copied into the new file.
Measured on the same fixture: the original JPEG carries the EXIF block; after one canvas re-encode at
quality 0.9, the output has no Exif marker and the inspector reports no tags. The re-encode
also changes the pixels slightly (it is lossy), which is why the file is smaller.
That behaviour is a feature for publishing and a hazard for archiving. Converting a photo to strip its location is a legitimate use; converting your only copy of a photo to “clean it up” loses the capture data permanently. Keep the original and work from a copy.
Where metadata goes when you upload
Platforms differ, and their behaviour changes without notice: some strip EXIF on upload, some keep it, and some keep it on the original but not on the resized version. The only reliable rule is to treat upload as disclosure. If the location must not be public, remove the metadata before the file leaves your device; do not rely on the destination to do it for you.
The same applies to screenshots and downloaded images. A screenshot has no camera EXIF because no camera was involved, while an image saved from a chat app may have had its metadata stripped by the app or may still carry it. There is no way to tell from the pixels, which is what the inspector is for.
Metadata is not only EXIF
Three other blocks travel with images and matter for different reasons. XMP is an XML metadata format that can carry editing history, keywords, ratings and rights information; Adobe tools write it, and it can survive operations that drop EXIF. IPTC carries captions, bylines and copyright fields used by news organisations. ICC profiles describe the colour space, and while they are not personal data, dropping them can change how the colours are rendered — a photo tagged with a wide-gamut profile can look dull once the profile is gone. The canvas re-encode described above drops all of them, which is why a stripped photo can look slightly different even before the JPEG quality loss is considered.
A practical routine before sharing
- Open the file in the EXIF viewer and read the Make, DateTime and any GPS indication.
- If the photo is going to a public place, re-encode it through the format converter or resizer — that removes the metadata as a side effect, along with the orientation flag.
- Check the result is upright; if the original relied on an orientation tag, the re-encode has already baked the rotation into the pixels.
- Keep the original if you might need the capture details later; the stripped copy is for sharing.
None of this is exotic. The metadata is the part of a photo that people forget is there, and removing it is a single local operation — no upload, no account, no server that keeps a copy.