Address #1. I am not saying either approach is authoritative. They serve as a point of discussion and examples of how the flexibility of the Web Annotation data structure allows you to make assertions through semantics.
The Annotation data must contain two types of datasets to properly georeference, one expressing points in pixels for warping the resource and another expressing the geocoordinates of where that warping applies in WGS84. Annotations may have multiple bodies or a Composite body, and both data structure semantics can make this statement.
For a Web Annotation that has multiple bodies, the Annotation should be interpreted as each Body is considered to be equally related to the target. That may be a bit off the mark or insufficient.
We might actually be trying to say the Items of each Body are related to each other, and both Items from both Bodies are required for the Annotation to be correctly interpreted. To do this, we could combine both datasets under a single Composite body. As you can see, this adds a bit more complexity to the Annotation but this complexity may be required for clients to understand exactly what this Annotation is trying to do.
This may require some input from other Annotation specialists, but both seem like prospective solutions to getting the pixel data out of the GeoJSON properties object.
Address #1. I am not saying either approach is authoritative. They serve as a point of discussion and examples of how the flexibility of the Web Annotation data structure allows you to make assertions through semantics.
The Annotation data must contain two types of datasets to properly georeference, one expressing points in pixels for warping the resource and another expressing the geocoordinates of where that warping applies in WGS84. Annotations may have multiple bodies or a Composite body, and both data structure semantics can make this statement.
For a Web Annotation that has multiple bodies, the Annotation should be interpreted as
each Body is considered to be equally related to the target. That may be a bit off the mark or insufficient.We might actually be trying to say
the Items of each Body are related to each other, and both Items from both Bodies are required for the Annotation to be correctly interpreted. To do this, we could combine both datasets under a single Composite body. As you can see, this adds a bit more complexity to the Annotation but this complexity may be required for clients to understand exactly what this Annotation is trying to do.This may require some input from other Annotation specialists, but both seem like prospective solutions to getting the
pixeldata out of the GeoJSON properties object.