The code in /contrib/packages/simulation/citybuild/object-edit uses cr rels to embed images in marked up text. In this case, the text is the content of a "sim_prop" or "sim_location" or "sim_character". An object can have up to 10 images associated with it (this arbitrary limit is set in the data model, like this:
select content_type__create_type(
'sim_location', -- content_type
'content_revision', -- supertype
'Location', -- pretty_name,
'Locations', -- pretty_plural
'sim_locations', -- table_name
'location_id', -- id_column
null -- name_method
);
select content_type__register_relation_type (
'sim_location', -- content_type
'image', -- target_type
'image', -- relation_tag
0, -- min_n
10 -- max_n
);
When you select an image from the dropdown (which shows all cr_items with type 'image'; type is set when it is uploaded, in this case using the same object-edit form but with image instead of location selected), you get an xml snippet that can be pasted into the body context of the sim_location. Whenever the sim_location is rendered, this xml snippet is converted to <img src="etc etc etc">.
You can install simulation on a HEAD server, go to citybuild, and try it yourself, or ping me for a login on a stable test server.
If this is going in the right direction for you, let me know and I can explain in more detail.
