Every photo on your computer is two things at once. There is the picture you can see, and there is a layer of text data attached to it that you usually cannot. That hidden text is photo metadata: the camera settings, the date, the location, and any titles, captions, keywords, and copyright information that you or your software have added. Open a JPEG in a viewer and you see the image. Open the same file in a metadata reader and you see a second story written in plain words.
Understanding this layer is the difference between a folder of anonymous files and a library you can search, license, attribute, and trust years from now. This guide explains what photo metadata actually is, breaks down the three formats you will keep running into (EXIF, IPTC, and XMP), shows which fields are worth your attention, and explains where the data lives and how to add it. Everything else in our metadata library builds on the foundation laid out here.
Metadata simply means "data about data." For a photograph, it is all the descriptive and technical information packaged alongside the pixels. Think of it as the label on a filing folder. The folder holds the document, but the label tells you what is inside without opening it. Your camera writes some of that label automatically. You write the rest, either by hand or with software.
A useful way to picture it: the image is the answer to "what does this look like," and the metadata answers everything else. When was it taken? What camera and lens? Where was the photographer standing? Who shot it, who owns it, and who can use it? What is in the frame, and what would you type to find it again in a library of fifty thousand photos? All of those answers ride along inside the file.
The reason this gets confusing is that "metadata" is not one thing stored in one place. It is several standards layered on top of each other, each created at a different time for a different purpose. Once you know the three main layers, the whole subject becomes simple.
Almost everything you will read about photo metadata comes down to three formats. They overlap a little, they coexist inside the same file, and they each have a clear job. Here is what separates them.
EXIF (Exchangeable Image File Format) is the technical layer, and it is written automatically the instant you press the shutter. You do not type any of it. Your camera or phone records the shooting conditions: shutter speed, aperture, ISO, focal length, lens model, camera body, white balance, flash status, and a date and time stamp. If location services are on, it also writes GPS coordinates. EXIF is factual and fixed. It describes how the photo was made, not what it means. For a full breakdown of every field and how to read it, see our EXIF data explained guide.
IPTC (named for the International Press Telecommunications Council, which created it for newswires) is the descriptive and rights layer. This is the part a human fills in. It holds the title, the caption or description, keywords, the creator's name, copyright notice, usage terms, location names, and contact details. Where EXIF tells a machine how the shot was exposed, IPTC tells a person and a search engine what the photo is about and who controls it. This is the layer that makes a photo findable, licensable, and properly credited.
XMP (Extensible Metadata Platform) is the newest of the three and works differently. It is not a separate set of fields so much as a flexible, modern wrapper that can store IPTC data, custom fields, edit history, ratings, color labels, and information from other standards, all in one structured format. Most current software reads and writes the IPTC fields through XMP behind the scenes. When you set keywords in Lightroom Classic, for example, they are stored in the XMP layer using the IPTC standard. Think of XMP as the shipping container and IPTC as one of the well organized crates inside it.
EXIF is what the camera knows, IPTC is what you tell it, and XMP is the modern container that carries your IPTC entries (plus extras) from one program to the next. Get those three roles straight and the rest of metadata is just detail.
Seeing the formats side by side makes their division of labor obvious. The table below shows who writes each layer, what it typically holds, and why it exists.
| Layer | Written by | Typical contents | Purpose |
|---|---|---|---|
| EXIF | Camera or phone, automatically | Shutter, aperture, ISO, focal length, lens, date, time, GPS | Record the technical conditions of capture |
| IPTC | You, or your software | Title, caption, keywords, creator, copyright, location names | Describe, credit, and license the image |
| XMP | Editing software | IPTC fields, ratings, labels, edit history, custom data | Carry everything between programs in one modern format |
In practice the layers are not rivals. A single image usually carries all three at once: EXIF from the camera, IPTC fields that you added, and an XMP block that organizes those IPTC fields plus whatever your editor wanted to remember. They sit together inside one file and a good metadata tool shows you all of them.
There are dozens of metadata fields, but a small handful do almost all the useful work. If you only ever fill in these, your photos will be searchable, creditable, and ready for almost any platform. These all live in the descriptive (IPTC) layer.
The title is a short name for the image, often one line. The caption, sometimes labeled "description," is a sentence or two explaining what is happening, where, and why it matters. Stock agencies, news systems, and content management systems all read these. A good caption is the single highest-value field for discoverability because it is full of natural language a search engine can index.
Keywords are the searchable tags that describe the subject, setting, mood, and concepts in the frame: "sunset," "coastline," "two people," "candid," "autumn." A well keyworded image might carry twenty to fifty terms. This is the field that decides whether your photo surfaces when someone searches a stock library or your own catalog. Choosing strong, consistent keywords is enough of a craft that we wrote a dedicated photo keywording guide on it.
The creator field names the photographer. The copyright field states ownership and usage terms, often something like "© Your Name. All rights reserved." Because this data is embedded in the file, it travels with the image when it is shared or downloaded, which is your strongest practical claim of ownership when a photo spreads beyond your control.
Alt text is a plain description of the image written for screen readers and for situations where the picture cannot load. The IPTC standard now includes a dedicated "Alt Text (Accessibility)" field so this description can travel with the file itself rather than living only in your website's HTML. Good alt text helps visually impaired visitors and gives search engines another clear, literal description of the content.
If filling in metadata feels overwhelming, commit to just five fields per image: title, caption, keywords, creator, and copyright. That set alone makes a photo findable, creditable, and ready for nearly any destination. Everything beyond it is refinement.
Metadata is stored in one of two places, and knowing which one matters because it changes what happens when you move, copy, or back up your photos.
For most formats, including JPEG, TIFF, PNG, and HEIC, the metadata is written directly inside the image file. The text and the pixels are one package. Copy that JPEG to a new drive, email it, or upload it, and the embedded metadata goes along for the ride. This is the simplest and safest arrangement: there is nothing to lose track of because there is only one file.
Raw camera files (such as Canon CR3, Nikon NEF, or Sony ARW) are a special case. Editors generally avoid writing into a proprietary raw file to keep the original untouched, so they store your metadata and edits in a small companion file that sits next to the photo with the same name and an .xmp extension. So IMG_4821.NEF gets a partner called IMG_4821.xmp. That companion is the sidecar, and it holds your keywords, ratings, copyright, and edit settings.
A sidecar only works if it stays with its photo. Move the raw file to a new folder and leave the .xmp behind, and all your metadata vanishes from that image. When you copy, archive, or back up raw files, you must move the .xmp files too. This single detail trips up a lot of people, which is why we cover it carefully in our guide to metadata for photo backups.
Metadata is invisible, so it is easy to ignore. But it quietly determines how useful your photos are over their entire life. There are four reasons it earns the small effort it takes.
Search engines, stock agencies, and your own catalog software find images by reading their titles, captions, and keywords. An image with rich metadata can be surfaced by a search; an image with none is effectively invisible, no matter how good the photo is.
Embedded creator and copyright fields travel with the file wherever it goes. When a photo is shared, scraped, or licensed, that data is your built-in claim of authorship and the easiest way for someone to credit you correctly.
Consistent keywords and captions turn a sprawling photo library into a database you can filter and sort. Finding every shot of a particular place, person, or theme becomes a one-second search instead of an afternoon of scrolling.
Decades from now, the only record of who, where, and when may be the metadata. A photo with a caption, date, and names stays meaningful long after the people who could explain it are gone. Without metadata, an old image is just an anonymous picture.
Here is the catch that surprises most people: embedding metadata does not guarantee it stays embedded. What happens to your data depends entirely on where you upload it. Some destinations preserve it, some read it and then discard it, and some strip it the moment the file lands.
Most large social networks re-encode images on upload to save space, and that process typically removes embedded EXIF and IPTC data. Some platforms read your caption or keywords first to inform their own systems, then deliver a stripped copy to viewers. The lesson: never rely on social uploads to carry your copyright or credit. Put your attribution in the visible caption too.
Stock and microstock libraries are the opposite. They read your embedded title, description, and keywords to power their search, and well filled IPTC fields directly affect whether your images get found and licensed. Here, metadata is not optional; it is the product's index.
On a site you control, you decide. Many content management systems preserve embedded metadata, and you can also map IPTC fields into on-page captions and alt text for accessibility and search. Self-hosting is the most reliable way to make sure your titles, credit, and descriptions stay intact and visible.
The practical takeaway is to keep a master copy of every image with full metadata embedded, and treat any uploaded version as potentially stripped. Your archive is the source of truth; the copies that go out into the world may arrive bare.
Writing metadata used to mean one thing: typing it into a panel, field by field, image by image. That approach still works and still matters, but it is no longer the only option.
Desktop applications like Adobe Lightroom Classic, Photo Mechanic, and Capture One let you fill in IPTC fields, build keyword lists, and apply shared values across a selection. ExifTool, a free command-line program, can read and write virtually any field in any format and is the power user's choice for automation. These tools give you total control, and for shared fields like copyright and location they are fast. The bottleneck is unique, per-image work: a distinct title, caption, and keyword set for every photo still has to be typed by a human, and that is what turns a large shoot into a long evening. If you want to weigh the options, our metadata tools comparison lays them out side by side.
The newer approach uses AI to look at each image and generate the descriptive fields for you. Instead of writing a caption and twenty keywords from a blank field, you review and refine what the AI produced. The model examines the actual content of the photo (the subjects, the setting, the colors and mood) and proposes a title, a caption, and a thorough keyword list in seconds. You stay in control, correcting anything wrong and adding details only you could know, like a client's name or the story behind the frame. The result is comprehensive metadata without the hours of typing.
For most people the best workflow blends the two: use a manual tool to stamp the fields shared across a whole shoot, then use an AI tool to generate the unique descriptive fields for each image. That combination is the heart of a fast, repeatable batch photo metadata workflow, and it scales from a dozen photos to thousands.
Photo metadata is not a niche technical topic. It is the difference between photos you can find, credit, and trust, and photos that are only as useful as your memory of where you left them. Once you understand the three layers (EXIF from the camera, IPTC from you, and XMP as the modern container that carries it), the field names stop being jargon and start being tools.
You do not have to do it all at once. Add five fields to your next batch of images, keep a fully tagged master copy of everything, and let the rest grow from there. Every photo you tag well is one you will be able to find, license, and pass on, long after you would otherwise have forgotten what it was.
Ready to go deeper on any layer? These guides build directly on the foundation above.
PhotoScanr analyzes your images and generates titles, captions, keywords, and descriptions for 7 platforms at once, including Lightroom Classic
Try PhotoScanr FreeFree to use . No sign-up required . Instant results