They could query based on the ["l", "picture-first"] label tag.

If they choose to show kind 1:s that don't have the "l" tag they'd need to client side filter by event / image format.