Metadata Cleaner ยท 6 min read
Beyond EXIF: XMP, IPTC, and the Hidden Metadata You Forgot About
EXIF gets all the attention, but a single JPEG can carry XMP, IPTC, ICC profiles, and Photoshop blocks too. Here is what each one stores and why stripping just EXIF is not enough.
Most metadata-cleaning tutorials stop at EXIF. That's a problem, because a modern JPEG straight out of Lightroom can carry four or five distinct metadata blocks layered on top of each other โ and stripping the EXIF section often leaves the rest untouched. If your goal is to actually anonymise an image, you need to know what else is hiding in there.
Why a JPEG Has More Than One Metadata Block
JPEG was designed to be extensible. The format allows arbitrary application-specific markers (APP0 through APP15), and over the years different vendors and standards bodies claimed different ones. EXIF lives in APP1, but so does XMP. IPTC sits inside Photoshop's APP13. ICC profiles take APP2. Each block has its own structure, its own parser, and its own privacy implications.
When you "remove metadata" with a half-decent tool, you might be stripping one block and leaving the others in place. Worse, some blocks duplicate information โ your GPS coordinates can appear in both EXIF and XMP, and removing only one of them leaves the location embedded.
XMP โ The One Adobe Built
XMP (Extensible Metadata Platform) is Adobe's XML-based metadata format. It was introduced in 2001 and is now part of ISO 16684. Anything edited in Lightroom, Photoshop, or Bridge writes XMP โ and XMP carries far more than EXIF ever did.
A typical XMP packet inside a Lightroom export contains:
- The full edit history โ every slider you touched, every crop, every brush stroke
- Star ratings, colour labels, and keywords you assigned
- Creator name, copyright notice, contact email, web URL
- The Lightroom catalogue identifier โ useful for stitching leaked images back to a specific photographer's library
- GPS coordinates (duplicated from EXIF)
XMP is human-readable XML, which makes it easy to inspect and easy to forget. Open a JPEG in a text editor and search for <x:xmpmeta>โ if it's there, that's your XMP block.
IPTC โ The One Newsrooms Built
IPTC (International Press Telecommunications Council) metadata predates XMP by a decade and is still standard in news and stock photography. It lives inside an APP13 Photoshop Image Resource Block, which is a quirky historical accident โ Photoshop became the dominant editor for press photos and its metadata container won by default.
IPTC fields include caption, byline, byline title, credit, source, headline, special instructions, country, city, location, and dozens more. Every wire-service photo has these populated. Even if you re-export the image, many tools preserve the IPTC block on the assumption that it's editorial intent rather than camera output.
ICC Profiles โ Less Personal, Still Identifying
An ICC colour profile describes how the image's colour values map to a real colour space. Most cameras embed sRGB or Adobe RGB. Most monitors embed a custom profile generated by colour calibration software, named after the device and sometimes the user (YashLaptop_2024-03-12.icc).
ICC profiles aren't usually privacy-sensitive on their own, but a custom profile name can fingerprint a device. If you're screenshotting from a calibrated workstation, the embedded profile might literally have your name in it.
Photoshop Image Resource Blocks
Beyond IPTC, Photoshop's APP13block can contain print settings, layer comp data, slice information, URLs assigned to image regions, and a thumbnail of the original image โ yes, a tiny embedded preview of what the photo looked like before you cropped or redacted it. This has bitten people. The classic case is "cropping" a sensitive document in Preview or a stock image editor and shipping a JPEG that still contains an uncropped 160-pixel thumbnail in its metadata.
What Else You Might Find
- MakerNotes โ proprietary EXIF extensions from Canon, Nikon, Sony, etc., often containing serial numbers, lens IDs, and shooting mode telemetry
- MPF (Multi-Picture Format) โ embedded preview images, common on smartphones
- Google Photos motion data โ Pixel phones embed a short MP4 of the moments before the still was captured, hidden in the file
- Apple Live Photo references โ link a HEIC still to its companion video file by UUID
The Practical Takeaway
If you're sharing an image and want it actually clean, don't trust a tool that promises to "remove EXIF." You need one that strips every metadata block โ EXIF, XMP, IPTC, MakerNotes, Photoshop IRBs, and any embedded thumbnails. Re-encoding the image (decoding to raw pixels and writing a fresh JPEG) is the surest way to drop everything. ExifTool's -all= flag does the same job for the technically inclined.
The browser-based metadata cleaner on this site re-encodes the pixels, which means there are no hidden blocks left behind. If you're using something else, run the output through ExifTool with -a -G1 afterwards to confirm โ and look for any output, not just EXIF.
References
- Adobe Systems Incorporated. (2020). XMP Specification Part 1: Data Model, Serialization, and Core Properties. Adobe.
- International Press Telecommunications Council. (2023). IPTC Photo Metadata Standard 2023.1. IPTC.
- CIPA. (2023). CIPA DC-008-2023: Exchangeable image file format for digital still cameras: Exif Version 3.0. Camera & Imaging Products Association.
- International Color Consortium. (2022). ICC.1:2022 โ Image technology colour management โ Architecture, profile format, and data structure. ICC.
- Harvey, P. (2024). ExifTool documentation โ supported metadata formats. ExifTool project.