The International Image Interoperability Framework, IIIF for short, is often presented as something for national libraries and large museums. It started there, and the biggest implementations are there. But the framework is just as useful to a small archive with a few thousand scanned photographs and one part-time technologist. The trick is to adopt the parts that pay off and skip the rest.
This guide explains what IIIF is, which pieces matter for a small collection, and how to get started without running a heavy image server.
What IIIF is, in plain terms
IIIF is a set of open specifications for delivering images and describing digital objects on the web. It is maintained by a community of institutions through the IIIF Consortium. The two core specifications are the Image API and the Presentation API, both at version 3.0 since 2020.
- The Image API defines a URL pattern for requesting an image or a part of it, at a given size, rotation and quality. A request looks like
/iiif/{identifier}/{region}/{size}/{rotation}/{quality}.{format}. An accompanyinginfo.jsonfile tells clients what sizes and tiles are available. - The Presentation API defines a JSON-LD document called a manifest. A manifest describes one object, such as a letter, a photograph album or a magazine issue, as a sequence of canvases with images and basic labels attached.
- Other APIs cover full-text search inside an object, access control, and change discovery for aggregators. They are useful, but most small archives can start without them.
The payoff comes from separation. Once your images and manifests follow the specifications, any compliant viewer can display them. Researchers can open your items in Mirador or the Universal Viewer, compare them side by side with items from other institutions, zoom deeply, and annotate. You do not have to build that interface yourself.
Why it matters for small collections
Small archives often hold material that is most valuable in context. A civil rights photograph means more beside the letters and songs of the same movement held somewhere else. The MetaScholar Initiative spent years pursuing that kind of cross-institution access with metadata harvesting. Its Music of Social Change project, for example, brought together records from libraries, museums and archives on music and the African American freedom struggle, including materials related to the topical song magazine described in the Broadside essay. The Highlander Folk School photographic history presented images one record page at a time, as most sites of its era did.
IIIF extends that idea from descriptions to the images themselves. Harvested metadata let scholars find an item across institutions. IIIF lets them look at it, closely, in one workspace, without downloading files or juggling browser tabs. For a small archive, that means its material can sit beside larger collections on equal terms.
Choosing a setup
There are three common ways to serve IIIF images. The right one depends on your volume, budget and staff.
| Approach | How it works | Good for | Watch out for |
|---|---|---|---|
| Static tiles | Pre-generate tiles and info.json with a tool such as libvips, then host them as ordinary files | Stable collections, low budgets, static hosting | Storage grows with tiles, and only pre-made sizes are available |
| Image server | Run software such as Cantaloupe or IIPImage that cuts tiles on demand from master files | Growing collections, flexible sizes | Needs a server, updates and some tuning |
| Hosted service | Use a platform or vendor that includes IIIF delivery | Teams without server staff | Ongoing fees and dependence on the provider |
For many small archives, static tiles are the best starting point. The Image API defines compliance levels, and level 0 is designed for exactly this case: a fixed set of pre-generated sizes and tiles served as files. It runs on cheap static hosting, has almost nothing to patch, and survives budget cuts well. You can move to an image server later without changing your identifiers if you plan the URL structure carefully.
Whichever route you choose, keep masters and access copies apart. Masters belong in preservation storage with checksums and replicated copies. Tiles and derivatives are disposable: if they are lost or a better tiling tool appears, you regenerate them from the masters. Treating access files as replaceable keeps preservation costs predictable and makes later migrations much less risky. It also means the IIIF layer can be rebuilt by a contractor or a future staff member from a short written procedure, which matters when the person who set it up moves on.
Getting started in five steps
- Pick a pilot collection. Choose fifty to two hundred items with clear rights and good existing metadata. Photographs or short manuscripts work well.
- Prepare master images. Keep high-quality TIFF or JPEG 2000 masters in preservation storage. Generate access derivatives from them, never the reverse.
- Generate tiles and info.json. Use a tool that writes the IIIF layout directly. Store the output under a stable path such as
/iiif/{identifier}/. - Write manifests from your metadata. Map title, date, creator, rights and a link back to the catalog record into each manifest. A short script can generate manifests from a spreadsheet or an OAI-PMH export.
- Test in two viewers. Open every manifest in at least two IIIF viewers and use a validator. Fix errors before you announce the URLs.
Then add a small viewer to your own item pages, and publish the manifest URL on each page so others can reuse it. That last step is what makes your collection interoperable rather than just zoomable.

Metadata and rights in manifests
A manifest is not a catalog record, and it should not try to be one. The Presentation API keeps descriptive metadata deliberately light: labels, a summary, a set of display pairs, a rights URI and a required statement for attribution. Detailed description belongs in your catalog, exposed in Dublin Core, MODS or another schema. Link to it from the manifest with the seeAlso or homepage properties.
Rights deserve care. The rights property expects a URI, typically from Creative Commons or RightsStatements.org. Use it consistently. Viewers and aggregators can then show users what they may do with an image, and you avoid publishing a high-resolution image of an item whose rights you have not reviewed. For sensitive material, consider serving only lower-resolution sizes, which static tiling makes easy to control.
What not to do yet
Small teams lose time by adopting too much at once. Defer authentication unless you have restricted material that must be online. Defer full-text search until you have transcriptions worth searching. Defer annotation servers until researchers ask for them. None of these are hard to add later, and all of them add maintenance.
What you should not defer is identifier planning. Decide how image and manifest URLs will look before you publish the first one, and commit to keeping them stable. Persistent, predictable URLs are the foundation every other IIIF feature builds on. Once they are in place, a small archive can join the same image ecosystem as the largest libraries, on its own terms and at its own pace. The specifications and a list of compliant software are at iiif.io.
