Where metadata lives inside an image file
A JPEG isn't one block of data — it's a series of named segments. Metadata lives in the header segments at the very front, before a single pixel of image data.
SOI — Start of Image APP0 — JFIF Header APP1 — EXIF Data Camera, GPS, settings, dates APP1 — XMP Data Keywords, tags, copyright, creator APP13 — IPTC Data SOF — Frame Header Image Data (pixels) Compressed pixel data · untouched by metadata writes EOI — End of Image Metadata Zone Written by exiftool. Pixels below are never touched. Inside APP1 — XMP Data (what we write) <x:xmpmeta xmlns:x="adobe:ns:meta/"> <rdf:RDF> <rdf:Description> <dc:subject> <rdf:Bag> <rdf:li>hr:pool-ramp</rdf:li> <rdf:li>golden-retriever</rdf:li> <rdf:li>large-dog</rdf:li> <rdf:li>pool</rdf:li> </rdf:Bag> ··· APP1 — EXIF (camera data) Make: Canon · Model: EOS R5 ISO: 400 · f/2.8 · 1/500s GPS: 41.8°N 87.6°W · Date: 2026-06-10 APP13 — IPTC Caption: "Golden lab on pool ramp" Creator: Photography Co. Copyright: © 2026 HandiProducts

The metadata segments (top colored blocks) are prepended to the file. The pixel data beneath is never modified. Tagging adds ~a few KB to file size — imperceptible.

How image metadata standards evolved
1985 – 1995
EXIF born in Japan
JEIDA (Japan Electronic Industries Development Association) developed EXIF to standardize how digital cameras store technical image data — shutter speed, aperture, ISO, white balance. Standardized as EXIF 2.1 in 1995. Every digital camera since has written EXIF data into every shot.
1991 – 1994
IPTC — the press wire standard
The International Press Telecommunications Council (IPTC) defined a metadata standard for news photos — caption, credit, keywords, copyright, location. AP, Reuters, and Getty Images adopted it to manage millions of wire photos. Still embedded in every JPEG alongside EXIF today.
2001
Adobe creates XMP
Adobe built XMP (Extensible Metadata Platform) to unify EXIF, IPTC, and custom metadata in a single open XML-based format. XMP can carry any metadata — keywords, color profiles, editing history, licensing, custom fields — in a structured, extensible way. Shipped with Photoshop CS.
2004
Adobe donates XMP to the public
Adobe submitted XMP to public standards bodies, enabling any software to implement it freely. Lightroom, Bridge, Capture One, and hundreds of other tools adopted it as the canonical way to store and exchange image metadata.
2012
XMP becomes ISO 16684
XMP was formally ratified as an international standard (ISO 16684-1). This cemented it as the open, vendor-neutral, permanent standard for embedded file metadata. No single company owns it. Any application can read and write it without licensing fees or lock-in.
Present
Ubiquitous — every camera, phone, and pro app
EXIF, IPTC, and XMP are embedded in virtually every photo taken by a digital camera or smartphone. Windows Explorer, macOS Spotlight, Apple Photos, Google Photos, Lightroom, Capture One, and all major CMSes read them natively. exiftool (the CLI we use) is the reference implementation — trusted by archivists, photographers, and media organizations worldwide.
Three layers, one file
Most image files carry all three simultaneously. They complement each other — each covers different ground.
EXIF

Exchangeable Image File Format

Technical camera data, written automatically by the device at capture. Read-only in most contexts — reflects the real capture conditions.

Contains: Camera make/model, shutter speed, aperture, ISO, focal length, GPS coordinates, timestamp, orientation, color profile.

📷 Who uses it: every camera, phone, GPS device, OS photo app
IPTC

International Press Telecommunications Council

Editorial and rights metadata. Created for news wire photos — meant for humans to write, not cameras to generate.

Contains: Caption, headline, keywords (legacy field), creator/photographer, copyright notice, city, country, category.

📰 Who uses it: AP, Reuters, Getty Images, stock photo agencies, news organizations
XMP

Extensible Metadata Platform

The modern open standard. XML-based, extensible to any schema. This is the field we write to — specifically dc:subject (Dublin Core keywords).

Contains: Any metadata you define — keywords/tags, edit history, licensing, color labels, star ratings, custom fields.

🔬 Who uses it: Adobe, Apple, Microsoft, Google, archivists, ISO 16684
Who depends on embedded metadata
XMP/EXIF keywords are not a niche feature — they underpin major industries and platforms.
📷

Professional Photography

Adobe Lightroom, Capture One, and Bridge are built around XMP. Every keyword, star rating, color label, and caption is stored as XMP. Industry standard for 20+ years.

📰

News & Wire Services

AP, Reuters, Getty Images, and AFP require IPTC/XMP metadata on every submitted photo. Caption, credit, and keyword fields are mandatory for photo syndication.

🖼️

Stock Photography

Shutterstock, Getty, iStock, and Adobe Stock use XMP keywords to power search across millions of images. Keywords = discoverability = revenue for photographers.

🏛️

Museums & Archives

The Library of Congress, Smithsonian, and national archives embed XMP into every digitized artifact. Ensures metadata survives for decades regardless of what database software exists.

⚕️

Medical Imaging

Medical imaging uses DICOM (a related standard) for embedded patient metadata. The same principle — critical data travels inside the file, not in a separate database.

🏠

Real Estate & MLS

MLS platforms read EXIF GPS data to auto-locate property photos. Many CMSes import image metadata directly into listing fields on upload.

🔎

Search Engines

Google reads EXIF and XMP data as image ranking signals. Alt text, IPTC captions, and copyright fields contribute to image search indexing.

⚖️

Legal & Copyright

Copyright holders embed ownership in XMP. Removal of copyright metadata is illegal in many jurisdictions. Courts use embedded metadata as evidence of ownership.

Where XMP keywords are readable and searchable
Platform / App Reads XMP Keywords Writes XMP Keywords Searchable by keyword Notes
Windows Explorer Reads Via exiftool ✅ Yes Details pane shows keywords; Windows Search indexes them
macOS Spotlight Reads Via exiftool ✅ Yes Cmd+Space search finds images by keyword
macOS Preview Reads No ❌ No Tools → Show Inspector shows all EXIF/XMP fields
Adobe Lightroom Read + Write Read + Write ✅ Yes Full keyword library; industry standard
Adobe Bridge Read + Write Read + Write ✅ Yes Free Adobe app; designed for metadata management
Capture One Read + Write Read + Write ✅ Yes Professional photo editing; full XMP support
Apple Photos Reads No ✅ Partial Imports EXIF; keywords visible but not easily searchable
Google Photos Reads No ❌ No Reads EXIF date/GPS; ignores XMP keywords
SharePoint Reads EXIF No ✅ Via columns Can surface EXIF in metadata columns; does not write back
WordPress Reads No ✅ Partial Imports caption, alt, title from IPTC/XMP on upload
exiftool (CLI) Read + Write Read + Write ✅ Yes The reference implementation; supports 25,000+ tags
digiKam (Linux/Win/Mac) Read + Write Read + Write ✅ Yes Open-source DAM; full XMP keyword management

🔑 Why this matters for our system

When exiftool writes a keyword like hr:pool-ramp into a JPEG, that tag is physically stored inside the file's XMP segment — an open ISO standard that has been in production for 25+ years. The tag is readable by Windows Explorer, macOS Spotlight, Adobe Lightroom, Bridge, and any XMP-aware CMS. It survives copy/paste, email, cloud sync, SharePoint, USB drives, and format migrations. No proprietary database, no vendor dependency, no subscription required. This is why we chose XMP over SharePoint-only columns or any other DAM-specific storage: the metadata belongs to the file, not to the platform.