EXIF vs IPTC vs XMP: The Three Kinds of Photo Metadata

EXIF, IPTC and XMP are the three main kinds of metadata stored inside photos. EXIF is technical information written automatically by the camera (device, settings, time, GPS); IPTC is descriptive and rights information added by people (caption, creator, copyright, keywords); XMP is a flexible XML format created by Adobe that can carry both, plus editing history and anything else an application wants to record. Most real-world photos contain more than one of them, often with overlapping values.

EXIF vs IPTC vs XMP at a glance

One naming trap: "IPTC" refers both to an organization's photo metadata standard (a list of fields) and to an older binary format called IIM. Today the same IPTC fields can be stored either in the legacy IIM block or inside XMP. When people say "IPTC data", they can mean either.

EXIF: the camera's technical record

EXIF was created in Japan in 1995 so that cameras, printers and software could exchange capture information. It reuses the tag structure of the TIFF format: each field has a fixed numeric ID and a binary value. A camera fills in what it knows, such as the shutter speed, aperture, ISO, focal length, flash, date and time, and, if it has a location fix, the GPS position and direction.

The tag list is defined by the standard, so EXIF cannot easily grow new kinds of information. Manufacturers get around this with the MakerNote tag, a proprietary block whose contents vary by brand and are not documented by the standard. Text fields were limited to ASCII until Exif 3.0 (2023) added UTF-8 support. For a full introduction, see what is EXIF data.

EXIF is the most privacy-sensitive of the three by default, because it is filled in automatically, without the photographer typing anything: GPS coordinates, the exact capture time, and on many cameras a body serial number.

IPTC: captions, credits and rights

The International Press Telecommunications Council, a consortium of news organizations, developed the Information Interchange Model (IIM) around 1990 so agencies could send photos with captions and credits attached. Adobe Photoshop later adopted a subset of those fields and stored them inside its own "image resource" block, which is why IPTC data in a JPEG still lives in a Photoshop segment today.

In 2004, IPTC worked with Adobe to define IPTC Core, which expresses the same fields in XMP. The current IPTC Photo Metadata Standard has two parts, Core and Extension, and is published on the IPTC website. Common fields include:

• Description (caption), Headline and Keywords.
• Creator (By-line), Credit Line, Copyright Notice and Rights Usage Terms.
• City, Sublocation, State/Province and Country: text locations typed by a person, separate from EXIF GPS.
• Creator's Contact Info: address, city, phone, email and website of the photographer.
• Digital Source Type: whether an image was captured by a camera, edited, or created by generative AI.

IPTC is mostly something you choose to add. But if you have ever filled in a contact template in a photo manager, every exported image may contain your email address and phone number.

XMP: Adobe's extensible container

Adobe released the Extensible Metadata Platform (XMP) in 2001, and its core was standardized as ISO 16684-1 in 2012. XMP is plain text: an XML/RDF packet organized into namespaces, each a set of fields. Common namespaces include:

• dc: (Dublin Core): title, creator, description, rights.
• xmp:: basic properties such as CreatorTool, CreateDate and ModifyDate.
• xmpMM: (media management): DocumentID, InstanceID, OriginalDocumentID and a History list of save and conversion steps with software names and timestamps.
• photoshop: and Iptc4xmpCore:: IPTC fields in XMP form, such as city and creator contact details.
• exif: and tiff:: copies of EXIF values, including GPS coordinates.
• crs:: Camera Raw and Lightroom develop settings.

Because any application can define a namespace, XMP is where software stores things the older standards never anticipated: phone depth maps and motion-photo information, editing recipes, and the document IDs that let software recognize that two files came from the same original. That last point matters for privacy: a shared image can carry an ID that links it to other images exported from the same source file.

For RAW photos, programs such as Lightroom often write XMP to a separate sidecar file instead of the image. Content Credentials (C2PA) are yet another, separate structure; they are covered in AI images and C2PA metadata.

Where each one is stored in JPEG, PNG and WebP

In WebP, the extended file header (VP8X) has flags that say whether EXIF, XMP and ICC chunks are present, as described in Google's WebP container specification. PNG also has free-form text chunks (tEXt, zTXt, iTXt) that some tools use for comments or generation parameters; see PNG metadata for the details.

HEIC, the default iPhone format, follows the same logic in a different container: EXIF and XMP are stored as separate metadata items alongside the image, which is why they survive when the photo is converted to JPEG during a transfer.

In all of these formats, the metadata blocks sit beside the compressed image data, not inside it. That is what makes lossless metadata removal possible: a tool can drop the APP1, APP13 or text chunks and copy the image data unchanged, with no recompression.

When the three overlap and disagree

Because the standards grew separately, the same information is often written several times:

• Copyright: EXIF Copyright, IPTC CopyrightNotice and XMP dc:rights.
• Author: EXIF Artist, IPTC By-line and XMP dc:creator; Windows may also write its own XPAuthor tag.
• Capture date: EXIF DateTimeOriginal, IPTC Date Created and XMP photoshop:DateCreated.
• GPS: EXIF GPS tags and XMP exif:GPSLatitude/exif:GPSLongitude.

The Metadata Working Group, an industry group whose members included Adobe, Apple, Canon, Microsoft, Nokia and Sony, published guidelines in 2008 on how software should reconcile these copies, but in practice files often disagree. The practical consequence is simple: removing one block does not remove the information. A tool that deletes EXIF GPS but leaves XMP can still leak your coordinates, and a viewer that shows "no location" may simply not be reading the XMP copy.

Removing EXIF, IPTC and XMP

For sharing, the safest choice is to remove all three at once. Our free metadata remover does that for JPEG, PNG and WebP files: it removes EXIF, XMP, IPTC, comments, text chunks and C2PA data in one pass, without re-encoding the image, and keeps only the ICC color profile and orientation flag. It removes everything, so it is not the right tool if you want to keep a copyright notice.

If you want to remove some groups and keep others, ExifTool lets you target each one:

# Remove only one family
exiftool -exif:all= photo.jpg
exiftool -xmp:all= photo.jpg
exiftool -iptc:all= photo.jpg

# Remove everything except the legacy IPTC block
exiftool -all= --iptc:all photo.jpg

# See which block each remaining value comes from
exiftool -a -G1 -s photo.jpg

Photographers who want to keep credit but drop location and serial numbers will find a balanced approach in which metadata photographers should keep. For a field-by-field list of everything these blocks can contain, see what metadata a photo contains.

The practical summary

• EXIF is automatic and technical; it is where GPS, capture time and serial numbers usually come from.
• IPTC is human-entered; check it for names, contact details and typed locations.
• XMP is the catch-all; it can duplicate EXIF and IPTC values and add edit history and document IDs.
• Inspect with a tool that shows groups (exiftool -G1), and remove all three before sharing files outside your control.