Tiled Deep Zoom and Open Seadragon

Manually entered Image API paramters is, of course, not how IIIF images are used in practice.

Once we start to see what support for the IIIF Image API can provide to software, we can unlock the power of the API.

With very high resolution images, for example, delivering the entire image to the user in their web browser is:

• slow
• a very inefficient use of network bandwidth
• demanding on the infrastructure that provides the image

The IIIF Image API, makes another way possible.

Tiled deep zoom of images.

Open Seadragon

Open Seadragon is an open source javascript component that can request image tiles, that is, just small fragments of the image, at a time, to create a smooth, zooming and panning user experience, without the need for Flash or plugins.

https://mcgrattan.org/osd.html?tilesource=https://iiif.bodleian.ox.ac.uk/iiif/image/446232db-9511-4e11-8b72-342e49dd4d82/info.json

If you use the developer tools in your browser, and look at what is happening when you zoom into the image, you'll see that the browser is only requesting just enough of the image at a time to support the panning and zooming user experience.

A typical tile might look like:

The tile size will depend on how the image server is configured, and the image server can tell (via the info.json) OpenSeadragon what size of tile request it prefers.

Using IIIF for delivery to services

These standard image requests don't just support the functionality of a viewer, and I will cover viewers in more detail later, but can also be used to deliver images directly to other software or services.

For example, Google, Microsoft, Amazon and others, provide services that can do image analysis, computer vision, and text extraction.

Azure:

https://azure.microsoft.com/en-us/services/cognitive-services/computer-vision/

I could paste in a IIIF Image URI into this demonstration page and extract the text.

With IIIF I could pass the whole page, or just pass a relevant section of the page.

https://www.nb.no/services/image/resolver/URN:NBN:no-nb_digifoto_20141030_00064_NB_WF_TVK_042510/full/1721,/0/default.jpg

More on images and text later.

Or for an image with no text:

https://www.nb.no/services/image/resolver/URN:NBN:no-nb_digifoto_20141209_00054_NB_MIT_FNR_20754/full/1802,/0/default.jpg

Or for a colour photograph from Harvard Art Museum:

https://ids.lib.harvard.edu/ids/iiif/10466656/full/full/0/native.jpg

We can pass this to automated tagging or image description.