How Image Metadata Works and What EXIF Can Reveal

Readable coordinates can expose where a photograph was captured. Camera model, lens, exposure, software and timestamps can add context. A coordinate does not by itself prove who took a photo, and missing metadata does not prove that no other copy or record exists.

Recognised fields are sorted into six groups: basic file information such as DateTimeOriginal and CreateDate; camera information including Make, Model, LensModel, FocalLength, FNumber, ExposureTime, ISO and Flash; location information covering latitude, longitude, GPSAltitude, GPSTimeStamp and GPSImgDirection; author and copyright fields such as Artist, Copyright, Credit, Source and UsageTerms; software and editing fields including Software, CreatorTool and processing History; and image information such as Orientation, resolution, ColorSpace and the embedded colour profile.

Reading every tag is unnecessary for most decisions. The analysis reduces a file to four yes-or-no privacy signals—location, camera, author, and software—and those four answer the question most people actually have.

Location is the obvious one, but note that GPSImgDirection records the compass bearing the camera was facing. Combined with coordinates, that describes a vantage point rather than merely a place. Author and copyright fields matter because they often carry a real name that was configured once in a camera or editor years ago and then forgotten. Software and editing history can reveal which application touched a file and when, which is sometimes more disclosing than the photograph.

Use Image Metadata when the question is what a file contains. It reads and reports only; it never writes a new image, and it accepts several files at once so you can triage a folder before sending any of it.

Use Remove Image Metadata when the question is give me a clean copy to share. It produces a new file and does not tell you what was there beforehand.

In practice you want both, in order: inspect the file you intend to send, strip it if the signals justify that, then inspect the cleaned download to confirm the signals are gone. Checking the output rather than trusting the operation is the step most people skip.

Inspection and removal do not cover the same formats. Three formats support both operations, and two support inspection only:

Because HEIC removal is unsupported, the route is to change format first. Converting a HEIC photo to JPG with the Image Converter decodes the pixels and re-encodes them, and that pixel pipeline does not carry HEIC metadata across—so the JPG it produces generally arrives without the original EXIF already.

Treat that as a likely outcome rather than a promise. Convert, then inspect the JPG and confirm the location and author signals are absent before sending it. If anything survived, the converted JPG is now a supported format for the remover.

Removal is not a surgical edit of a metadata block. The image is decoded, drawn onto a canvas, and re-encoded from those pixels. Nothing survives that round trip: EXIF, XMP, IPTC and the embedded colour profile are all absent from the result, because the new file is built from pixel data alone.

That has three consequences worth knowing. A JPG is re-encoded at quality 0.95, which is lossy—visually close, but not byte-identical to the original—while a PNG is re-encoded losslessly. The embedded colour profile is dropped, so a wide-gamut photo can shift slightly on a colour-managed display. And because browsers apply EXIF orientation while decoding, the canvas is already upright, so the cleaned copy needs no orientation tag and should not appear rotated.

Verification is a separate step from cleaning, and it takes about fifteen seconds.

The cleaned download is named after the file it came from, with a suffix appended—so sending-photo.jpg becomes sending-photo-metadata-removed.jpg. Two things follow. The original basename is carried straight through, so if it contained a person's name, a case reference or a document type, stripping the EXIF did nothing about it. And the suffix itself announces that the file was deliberately scrubbed, which is occasionally information you would rather not volunteer.

Rename the file before sending it. If you need a record of what the original contained—for a compliance trail, or simply to know what you removed—export the analysis as text from the inspector first, because after the clean copy exists there is nothing left in it to read.

A cleaned download does not erase the original, cloud backups, caches, message copies, filenames, or information another service adds after you upload. Vendor-specific or unusual tags may not be readable in the first place, so an empty report is weaker evidence than a report that lists fields and shows them gone afterwards.

Metadata is also not the only thing a photograph discloses. Street signs, house numbers, reflections, distinctive skylines and visible screens survive every metadata operation, because they are pixels. Inspection covers up to 20 files per batch, 80 MB per file and 60 megapixels per image; larger files are rejected rather than partially analysed.