EXIF vs IPTC vs XMP: What’s the Difference?

EXIF stores mainly camera and capture information, IPTC focuses on descriptive and rights metadata, while XMP provides an extensible framework that can store and synchronize metadata across applications.

Quick Answer

EXIF, IPTC, and XMP are three coexisting metadata standards embedded inside the same image file. EXIF stores camera-generated technical data (shutter speed, GPS coordinates, lens model). IPTC stores human-written editorial data (caption, keywords, copyright). XMP is Adobe's modern XML wrapper that mirrors both and adds extensions for ratings, edit history, and rights management. EXIF rules privacy. IPTC rules stock photo sales and Google Images visibility. XMP rules professional workflows in Lightroom, Photoshop, and Bridge. Use all three intentionally, or you're losing visibility, control, or both.

Quick comparison table

Now let's go layer by layer.

What is EXIF? The camera-generated layer

EXIF stands for Exchangeable Image File Format. CIPA, the Camera & Imaging Products Association based in Japan, finalized the modern specification in the mid-1990s, and every modern smartphone and digital camera writes it automatically the moment the shutter fires. A typical photo from an iPhone 16 or a Sony A7 IV embeds 30 to 50 EXIF fields without you doing anything: shutter speed, aperture, ISO, focal length, lens model, camera serial number, software version, white balance, exposure compensation, and the big one, GPS coordinates accurate to roughly three meters.

EXIF is binary. It lives inside a specific marker segment of the JPEG file and is read directly by Windows File Explorer, macOS Finder, ExifTool by Phil Harvey, and Google Photos. If you've never inspected what a single photo actually contains, our primer on what EXIF data really holds walks through every standard field with screenshots. Most users are surprised by the GPS line. Some are alarmed by the camera serial number, which can theoretically link multiple photos back to the same physical device years apart.

What EXIF is not designed for: keywords, captions, or anything you'd type yourself. The specification includes a few user-comment fields, but stock platforms and Google Images do not parse them as ranking signals. EXIF is the camera's diary, not the photographer's.

Best for: forensic photo verification, lens and exposure debugging, copyright proof of capture date, and, when you don't want it, the single biggest privacy leak in casual photo sharing. A geotagged restaurant pic posted to a public profile reveals your exact GPS location to anyone with a free EXIF viewer. That is why dedicated EXIF removal exists in the first place. For a structured introduction to the spec, the canonical reference is our what is EXIF guide.

What is IPTC? The editorial layer

IPTC came first. The International Press Telecommunications Council, headquartered in London, defined the IIM (Information Interchange Model) in 1979, sixteen years before EXIF existed. Its job was simple: let news agencies syndicate photos with structured caption, byline, copyright, and keyword data attached. Reuters, Associated Press, and Agence France-Presse use IPTC daily. So does every major stock platform on Earth.

IPTC is human-written. The fields most photographers care about are:

1. Description / Caption — what's in the image
2. Keywords — typically 25 to 50 tags per image; this is where stock photo discoverability is won or lost
3. Headline — short title
4. Creator / Byline — your name
5. Copyright Notice — required by Shutterstock, Adobe Stock, Getty Images, Alamy, and 500px
6. Subject Code — IPTC's controlled vocabulary
7. Location — written, not GPS-numeric (e.g. "Marrakech, Morocco")

A complete reference of every IPTC field, including the legacy IIM names and the modern Core/Extension schema, lives in our IPTC metadata fields breakdown.

If you sell on stock sites, IPTC is your revenue layer. Shutterstock indexes the Description and Keywords fields for in-platform search ranking. Adobe Stock cross-references your Title, Keywords, and Description when ranking results. Getty Images applies its own controlled vocabulary that maps to IPTC fields. Empty those fields and your photo is invisible, no matter how good the shot is.

Best for: stock photo sales, news photography licensing, Google Images ranking signals, copyright assertion, and any image that needs to be searchable by topic. Generating 50 high-relevance IPTC keywords for one image used to take 8 to 12 minutes manually. An AI IPTC keyword generator compresses that into seconds while keeping the keywords aligned with each platform's preferred phrasing.

What is XMP? The Adobe wrapper that ties it all together

XMP, the Extensible Metadata Platform, was published by Adobe in 2001 and re-released as an ISO standard in 2012. Unlike EXIF (binary) and IPTC (binary IIM headers), XMP is XML built on RDF. Human-readable. Schema-extensible. Embeddable in JPEG, PNG, TIFF, PSD, PDF, and even MP4.

The clever part: XMP can mirror EXIF and IPTC fields, plus carry its own. When Adobe Lightroom imports a photo, it reads EXIF, reads IPTC, and writes a unified XMP block that includes everything plus Lightroom-specific data. Star ratings. Color labels. Develop module adjustments. Virtual copy parents. Edit history. Adobe Photoshop does the same. Adobe Bridge is essentially an XMP browser.

Seven things XMP does that EXIF and IPTC cannot:

1. Carry edit history (which sliders moved, when, by whom)
2. Store custom schemas (rights management, accessibility, AI-training opt-out flags)
3. Embed in non-image formats (PDF, MP4, AI files)
4. Be written as a sidecar .xmp file when the image format doesn't allow embedding, or when you don't want to alter the original raw
5. Carry hierarchical keywords (Travel > Asia > Japan > Tokyo) which Lightroom uses for collection logic
6. Survive Lightroom-to-Photoshop round trips without data loss
7. Hold the IPTC Extension fields (model release, property release, AI-generated flag added by IPTC Council in recent updates)

Most modern stock platforms read XMP first, IPTC second, and EXIF mostly for date and GPS. If you write keywords inside Lightroom, they end up in XMP. Many tools then sync them back into the IPTC-IIM block on export so older systems can still read them. If you've never inspected what a stack of XMP, IPTC, and EXIF actually looks like inside a real file, our examples of metadata walks through real XMP packets line by line.

How EXIF, IPTC, and XMP coexist inside one JPEG

Open any JPEG in a hex editor and you'll see all three formats stacked inside dedicated marker segments. EXIF sits in the APP1 marker. The legacy IPTC-IIM block lives in the APP13 marker, also called the Photoshop image resource block, because Photoshop popularized it. XMP gets its own APP1 marker right after EXIF, identified by an http://ns.adobe.com/xap/1.0/ namespace string.

This means a single JPEG file can simultaneously contain a GPS coordinate (in EXIF), a handwritten caption (in IPTC-IIM), and the same caption mirrored in XMP plus Lightroom edit history, hierarchical keywords, and a star rating (in XMP). Three layers, one file, three different parsers reading three different things.

When you edit metadata in a modern tool, you should always write to all three layers, or at minimum to XMP and IPTC-IIM, which most pro tools sync automatically. Why? Because not every downstream system reads every layer. Shutterstock reads IPTC. Lightroom reads XMP. Old PHP image scrapers from 2008 only read EXIF. Write into one layer alone and you'll wonder why your caption appears in some places and disappears from others.

This is where dedicated metadata injectors earn their keep. A good bulk metadata injector writes coherently across EXIF, IPTC, and XMP simultaneously, so every platform that touches the file sees the same data. Tools that write to one layer leave your file inconsistent across readers, which produces the most frustrating support tickets in stock photography: "It looks fine in Lightroom but Shutterstock rejected my keywords."

Which standard ranks on Google Images in 2026?

Google has never published a complete list of the metadata fields it parses for image ranking. Based on Google Search Central documentation, structured-data updates, and observed indexing behavior, the picture is reasonably clear:

1. EXIF: Google reads DateTimeOriginal (publish-date hints), GPS data (occasionally for "near me" queries), and camera model (rarely meaningful for ranking).
2. IPTC: This is the heavy hitter. Caption, Description, Keywords, and Creator are all parseable signals. Google's Image License Metadata feature explicitly relies on IPTC fields: Creator, Credit, Copyright, and Web Statement of Rights.
3. XMP: Google reads XMP-DC (Dublin Core), XMP-rights, and the IPTC fields mirrored into XMP.

The takeaway: filling out IPTC properly, and letting your tool sync those values into XMP, moves the needle for Google Images far more than tinkering with EXIF. The full ranking framework, which combines filename, alt text, surrounding text, schema markup, and embedded IPTC, is detailed in our image metadata for better ranking guide. It also explains why image filename optimization alone never closes the gap; the embedded layer matters more than most SEOs realize.

Which standard matters for stock photo sales?

If your goal is selling images on Shutterstock, Adobe Stock, Getty Images, Alamy, EyeEm, or 500px, the priority order changes:

1. IPTC is non-negotiable. Title, Description, and Keywords drive the majority of in-platform search ranking. Empty fields equal invisible images.
2. XMP matters when you upload from Lightroom, because Lightroom-native keywords end up in XMP. Most platforms parse this correctly, but always confirm by re-reading the file after export with a separate tool.
3. EXIF mostly matters for DateTimeOriginal, so the platform can sort recent uploads, and sometimes for buyer trust signals (lens used, camera body).

The keyword field deserves its own paragraph. Shutterstock allows up to 50 IPTC keywords per image. Adobe Stock allows the same upper bound. Getty applies tighter editorial review but still rewards 30+ relevant keywords. Filling each slot manually takes 8 to 12 minutes per image. A photographer with 500 unprocessed images is looking at roughly 60 to 100 hours of pure typing. AI keyword generation collapses that to seconds per image with comparable relevance, provided the underlying model has been trained on stock-platform language patterns rather than on generic image-captioning datasets.

Common mistakes when editing across all three layers

Five mistakes I see constantly when auditing photographer workflows:

1. Editing IPTC in one tool, EXIF in another, and XMP nowhere. The file ends up with conflicting metadata across layers, and downstream platforms grab whichever layer they read first. Result: chaotic search ranking.
2. Stripping EXIF for privacy without preserving IPTC. Many privacy tools nuke all metadata, which deletes your IPTC copyright and keywords too. Use a tool that strips selectively rather than scorching every layer.
3. Using Photoshop's File > File Info dialog only. It writes to XMP and IPTC-IIM but does not always touch every EXIF field. Combine with a dedicated EXIF utility for full coverage.
4. Trusting Instagram or Facebook to preserve any of it. They strip metadata on upload by design. If you need credit to survive social sharing, use visible watermarks plus C2PA content credentials.
5. Bulk writing the same keyword set to 200 different photos. Stock platforms penalize keyword stuffing and identical-keyword spam. Each image needs its own relevant set, not a copy-paste template.

Information gain: the metadata stack most photographers never document

Here's the part competitors leave out. Inside one professional JPEG you can find more than three layers if you look closely:

1. EXIF APP1 segment (camera data)
2. XMP APP1 segment (Adobe namespace, mirrors IPTC and adds Lightroom data)
3. Photoshop APP13 / IRB block (legacy IPTC-IIM)
4. Optional ICC color profile (display rendering, often confused with metadata but technically separate)
5. Optional embedded thumbnail in EXIF
6. Optional MakerNote (proprietary camera-vendor data; Canon, Nikon, and Sony each define their own internal format inside this segment)

When a tool claims to "remove all metadata," it must address every one of these segments. When a tool claims to "inject metadata," it should write to APP1 (XMP) and APP13 (IPTC-IIM) simultaneously, while letting the photographer choose what stays in EXIF. Tools that touch only one segment leave your file inconsistent. This is why testing across multiple readers, namely Windows Properties, Adobe Bridge, ExifTool, and a dedicated stock-platform preview, matters in practice. A perfectly clean Bridge view does not guarantee a clean Shutterstock submission, because the two readers may be looking at different segments.

FAQ

What's the simplest difference between EXIF, IPTC, and XMP? EXIF is what your camera writes automatically (technical settings, GPS). IPTC is what you write manually for editorial use (caption, keywords, copyright). XMP is Adobe's modern wrapper that can hold both plus extensions like edit history. Camera equals EXIF. You equal IPTC. Adobe equals XMP.

Do I need to edit all three or just one? If you sell stock photos or want to rank in Google Images, edit IPTC at minimum. Most modern tools auto-sync your IPTC edits into XMP. Leave EXIF alone unless you're stripping GPS for privacy or correcting a wrong capture date.

Does Instagram preserve EXIF, IPTC, or XMP? Instagram strips most metadata on upload, including EXIF GPS, IPTC keywords, and XMP. A few EXIF fields like camera model occasionally survive. Treat any social platform upload as a metadata-loss event.

Which standard does Google Images use for ranking? Primarily IPTC fields parsed via the Image License Metadata framework: Creator, Credit, Copyright, Caption, and Keywords. XMP is read when it mirrors IPTC values. EXIF rarely affects ranking but can affect indexing freshness via DateTimeOriginal.

Can XMP replace EXIF and IPTC entirely? Technically XMP can carry every field EXIF and IPTC define. In practice, older systems (legacy CMSes, basic image scrapers, some stock-platform pipelines) still read the binary EXIF and IPTC-IIM blocks first. Best practice: write all three coherently rather than relying on XMP alone.

How many IPTC keywords should I add per image? Stock platforms allow up to 50 keywords (Shutterstock, Adobe Stock). Aim for 25 to 40 highly relevant ones. Beyond 40 you risk diminishing returns and keyword-stuffing penalties. Hierarchical keywords stored in XMP can help your Lightroom workflow without inflating the visible IPTC count.

Are EXIF, IPTC, and XMP supported in PNG and WebP? EXIF is supported in WebP (since the format added it in 2017) and partially in PNG via custom chunks. IPTC-IIM is rare outside JPEG and TIFF. XMP works in PNG, WebP, TIFF, PSD, and PDF. JPEG remains the gold standard for full three-layer metadata support.

The bottom line

Three standards, one file, three different jobs. EXIF protects (or exposes) your privacy. IPTC drives your stock photo income and your Google Images ranking. XMP keeps your professional workflow consistent across every Adobe app and most modern tools.

The mistake is treating them as interchangeable. They aren't. The other mistake is editing them in three different apps that don't sync. A coherent metadata stack, written once and written everywhere it needs to live, is the difference between an image that sells and one that quietly disappears from search results.

Want to write EXIF, IPTC, and XMP coherently across thousands of photos in a single pass? See pricing and plans.