Annotation with a Non-Rectangular Polygon

Use Case

For a IIIF resource, you would like to add a simple annotation to the resource, where the area of the resource you would like to highlight is not a rectangle. The shape of the focus area needs to be delineated with precision to highlight only the portion of the image that is relevant to the annotation.

Implementation Notes

The IIIF Presentation 3.0 API does not itself discuss non-rectangular annotations, incorporating them from the W3C Web Annotation Data Model by reference. For a full description of this and other web annotations used in IIIF annotations, we recommend you read that document.

The W3C data model requires non-rectangular polygonal annotations to be described in the Scalable Vector Graphic (SVG) markup format. Note that client parsing of SVG varies, and valid SVG may not display as you expect.

When reviewing your SVG data, remove all styling and transformation features, per the W3C data model. To ensure your SVG markup is valid, you can use the W3C validator. SVG can be valid markup as absolute points in a coordinate space or as relative points along a path.

Sizing and placement of the SVG polygon in relation to its target takes some special attention. If the target is the entire Canvas, then the SVG viewport is assumed to cover the entire Canvas. On the other hand, if the SVG’s coordinates are mapped to a part of a Canvas, then its target must be the rectangular bounding box of the Canvas in which the SVG viewport should be placed. If the dimensions of the viewport and its target region (either the bounding box or the entire Canvas) are not the same, then the SVG must be scaled such that it covers the region. This may result in different scaling ratios for the X and Y dimensions.


This approach should not be used to describe non-rotated rectangular regions.


In this Manifest, we are highlighting a fountain with a statue on top of it in Göttingen, and imagining that we want to be fairly precise in our highlight. The client should not show the bounding box on the image.

JSON-LD | View in Mirador