What is IIIF (and how to get manuscript images out of a IIIF viewer)

IIIF, the International Image Interoperability Framework, pronounced "triple-eye-eff", defines interoperable APIs for image delivery and digital-object presentation. A repository may expose a manifest, an Image API service, or both.

IIIF resources can connect a repository's images and descriptive structure to compatible viewers and research tools.

What IIIF is, in plain language

The IIIF specifications give repositories and software a common way to deliver images and describe digital objects. Compatible viewers can use resources published by different institutions without requiring one proprietary viewer format.

The Image API 3.0 defines a URL grammar for requesting an image or region at a chosen size. The Presentation API 3.0 defines manifests and related presentation resources. Version 2 resources remain in use, so read the version and profile supplied by the service.

Recognizing IIIF in the wild

Smooth deep zoom can suggest IIIF, but it does not prove that a site uses IIIF or exposes a manifest. Look for a IIIF logo, an “IIIF manifest” link, or an entry in an information or share panel. Mirador and Universal Viewer consume IIIF resources; OpenSeadragon supplies tiled-image viewing but does not by itself establish IIIF support.

The manifest: a machine-readable facsimile

A manifest is a JSON-LD representation of an object. It can provide labels, repository information, rights statements, behavior, and an ordered sequence of canvases. A canvas is an abstract space on which image and annotation resources are placed; depending on the institution, one canvas may represent a page, an opening, or another view. A manifest describes the digital presentation. It is not the manuscript or a preservation master.

Open a manifest URL in your browser and you get raw JSON. It is less friendly than a viewer but perfectly legible: scroll and you will find the canvases, their labels, and inside each one an image service URL. That URL is the other half of the trick.

Getting a single image

The Image API defines every image request as a URL with a fixed anatomy:

The parameters apply in that order: crop a region, scale it, rotate it, pick a quality, pick a format. To ask for the entire image at the largest size the server will give, request:

• Version 3: .../identifier/full/max/0/default.jpg
• Version 2: inspect info.json and the service profile. full is the level-0 size syntax; max exists in Image API 2.1 but is an optional feature.

A request can specify the full image, maximum permitted size, zero rotation, default quality, and JPEG format. The permitted maximum may be smaller than the preservation master.

Check the viewer's download control first. If the repository exposes an image service, open its info.json and follow the advertised Image API version, profile, sizes, formats, authentication requirements, and limits. The Image API 2.1 compliance table shows which features a client may assume and which remain optional. A maximum-size request returns the largest derivative the service permits, which may be smaller than the preservation master or unsuitable for publication.

Getting a whole manuscript

For a whole manuscript, use the repository's bulk-download option when one exists. Otherwise, a script or IIIF-aware tool can read the manifest and request the derivatives that the image service advertises. The IIIF community's software directory lists current tools. Do not use tile endpoints or third-party downloaders to evade a repository's size limit, authentication, access control, or stated terms.

Rights and manners

An available image is not permission to reuse it. Read the manifest's rights statement and the repository's terms; study access may coexist with separate publication conditions. The guide on tracking reproduction rights explains how to retain that evidence. Respect published rate limits and bulk-download instructions. Reuse local copies instead of sending repeated high-resolution requests.

After the download: the filename problem

A bulk download may use service identifiers such as b8447128-0044.jpg rather than folio labels. Preserve the manifest and rename working copies with the verified foliation; see renaming photographs to folio numbers and organizing manuscript research photos.

The directory of digitized manuscript collections and reuse terms links to repository policies for the collections listed there.