Skip to content


Image focal point hint

This is a quick idea that may already exist somewhere. Let me know if “there’s a vocab for that”.

We have a large set of images of our university buildings. There’s a variety of sizes & aspect ratios. Sometimes there’s more than one image of a building.

To render these in the university templates we need to trim them to certain sizes and aspect ratios. What would be useful is if we could store a “hint” of where the most important content is in the picture. For example take this image:

tinkering

Clearly the most important part of this picture is the relationship between the researcher and the tool. I would say about here:

tinker with X

Which is about 36% from the left, and 50% from the bottom. What I’m wondering is if there is (or should be) some standard terms for indicating the focal point of this picture eg.

<http://example.com/pictures/research.jpg> ns:focalPointX "0.36"^xsd:float .
<http://example.com/pictures/research.jpg> ns:focalPointY "0.5"^xsd:float .

That way our HTML page generation can get cropped images but instead of a default focus (usually the centre point) it could know how to crop for this picture. You can see the results of making a portrait crop of this image using a focal point hint and without:

Without hint:

pic4

Using hint:

pic3

I think this makes a massive difference and seems like a really useful thing to optionally store with our images and publish as part of the metadata for them in the open data service.

Other useful metatags for images linked in open data would be:

  • An indication that this image should be treated as the illustrative image of something.
  • For logos; if the image is more suited for a light or dark background (ideally letting the renderer pick the more suitable variation
  • For background style images, a hint if the image is a light or dark background (so we know how to place contrasting text over the top.)

Does this exist already somewhere out in ontology land? Are there any other useful things we should add?

Posted in Best Practice, RDF.


One Response

Stay in touch with the conversation, subscribe to the RSS feed for comments on this post.

  1. Alexander Dutton says

    When asked, Dan Brickley told me ages ago that it seemed reasonable to use foaf:img to annotate a thing with its illustrative image.

    You could use the Web Annotation vocabulary to annotate bits of an image (see e.g. https://www.w3.org/TR/annotation-vocab/#svgselector), but you’d then have to say what it is that’s the focal point (or have a blank node of rdf:type ex:FocalPoint?). Might be worthwhile if you’re labelling all the buildings in an image though. I also can’t work out if the building should be the target or the body of the annotation :/



Some HTML is OK

or, reply to this post via trackback.