Targeting specific 3D regions of a scene

[tomcrane]

Split from #2255

A comment might target a 3D volume within the Scene.

A suggested approach is to use a WktSelector:

https://en.wikipedia.org/wiki/Well-known_text_representation_of_geometry

    "target": [
        {
            "type": "SpecificResource",
            "source": [
                {
                    "id": "https://example.org/iiif/scene1/page/p1/1",
                    "type": "Scene"
                }
            ],

       "selector": [
           {
               "type": "WktSelector",
               "value": "POLYGONZ(( 0 0 0, 1 0 0, 0 1 0, 0 0 1 ))"
           }
        ]
        }
    ]

Note that the value uses POLYGONZ not POLYGON: https://gis.stackexchange.com/a/206631

[azaroth42]

Other possible approaches could include:

• Sphere, given a central point and a radius (and impossible with POLYGONZ)
• Cuboid, given two points (top front left, bottom back right)
• Cuboid, given five points (x,y,w,h,+depth)

POLYGONZ seems like the most flexible and most standardized. We should determine if Sphere is a use case or not.

[azaroth42]

I think this is ready for eds, although not discussed directly, the only concerns were around arbitrary segmentation of geometries

[glenrobson]

We discussed this in the European timed 3D update and there were concerns raised about the complexity of working out the coordinates required for a POLYGONZ target. Particularly if you are painting hundreds of canvases into a scene.

Reading the spec at the moment it is easy to read that this is the only way to paint canvases into scenes and I think the more common case of painting a canvas into a scene and then using transforming and scaling is lost as the only example is the PLOYGONZ. It may help to include the diagram mentioned in the text and possible a subheading for the POLYGONZ mentioning its only required if you want to skew the image.

From further disucssion in the meeting it seemed like POLYGONZ option is an edge case and there are questions on how oftern a user would want to preform these skewing on images. It might be useful to include a use case for when you might want to do this.