Exif viewer: every tag, named and decoded

Directories of tags

Exif is a small TIFF file stored inside the photo. It is built of directories, and each directory is a list of tags: a number, a type, a count and a value. This viewer walks every directory it finds and prints every tag in it — nothing is summarised away. Open a card’s What it says and then Every tag, raw: each directory is a table of its own, with the tag’s number, its registered name, its type and its value in words.

The viewer names 310 tags: 273 from the TIFF, Exif and DNG registers, 32 GPS tags and 5 interoperability tags. A tag it has no name for is still shown, as an unknown tag with its number and its raw value. 38 tags have a table of meanings, so Orientation 6 reads as a photo on its side and metering mode 5 as a pattern. Exposure time is written as a fraction of a second, aperture as an f-number, and the APEX values are converted. The same tags are also filed in plain words, by what they give away: where, when, the camera and lens, the exposure, people and software.

GPS: coordinates, and a warning

If the file holds a complete position, the card turns red and says so first, as decimal degrees and as degrees, minutes and seconds, with the altitude, the accuracy of the fix, the direction the lens faced and the GPS time where they are there. You can copy the coordinates with one press. There is no map on this page and no link to one: a map would have to fetch tiles for that spot from a map server, and that request would tell the server where your photo was taken. To take the position out, strip GPS from the photo — all the metadata, or the location only.

Beyond Exif

XMP is an XML packet. The viewer lists the properties it finds and shows the packet exactly as it is stored. IPTC fields — caption, credit, city, keywords — live in a Photoshop resource block, and each dataset is listed by its record and number; 39 dataset names are known. The colour profile is read by its header: its description, its version, its class and its colour space. In a PNG the viewer reads tEXt, zTXt and iTXt text, eXIf, pHYs, iCCP and tIME; in a WebP, the VP8X flags and the EXIF, XMP and ICCP chunks; in HEIC and AVIF, the brands, the items, the rotation property and the Exif and XMP items. The last table is the file itself, part by part: every segment or chunk with its offset, its size, and whether a strip would remove it.

What it does not do

The maker note is a block in the camera maker’s own format. The viewer names it, measures it, shows the first characters when they are readable, and does not decode it. JIS-encoded comments are named, not decoded. Extended XMP is named and measured only. HEIC and AVIF items that point into other items are not followed. Where a value points outside its block, or a directory loops back on itself, the viewer says so and reads no further down that path. Every value is written as text: a tag that holds markup shows the markup, and nothing in a file becomes part of this page.

Where the names come from

From your file and nowhere else. The tag names were generated from Pillow 11.3.0’s register; the meanings of values were written for this bench and checked, key by key, against ExifTool’s published tables.

How to check that nothing is uploaded. The page’s own scripts make one request of their own — the sample card, from this site, when you press for it — and none that carries a file. The folder is served with a Content-Security-Policy header that lets the page connect only to this site and to Google’s analytics hosts, so even a mistake here could not send a photo anywhere else: the browser would refuse it. Open the developer tools, choose the Network tab, clear it, then add a photo and run the tool, and watch. Google’s analytics counts the visit and that a tool was used; nothing in this page hands it a file name, a pixel or a tag.