You could make a heightmap (ortho render with just illumination of altitude, as I described in another thread), and import in PS. Then you'd have a basis to work from, and have an idea about hills and such.
Then make a a layer for each building type, and paint in the mask. If you need very small points to fill with specific objects, you'd need a pretty large map.
Then copy three layers to a new RGB file and copy each layer into a channel. name the file accordingly.
Import in TG, project either by that same ortho camera, or plan Y and set size and location so it fits, then split (convert-blue to scalar, etc), and use for the pops.