I've been working on my next report version, and want to use my virtual landscape for illustrations. I wasn't satisfied with the level of accuracy in tree placement. I had located trees and shrubs using aerial photos paired with perspective shots of the topo map, but the process seemed too loose.
I have ten usable photos of the scene, five aerial and five on the ground. Over the last several weeks I've positioned each one over my virtual terrain, paired with a virtual camera to recreate the viewpoint of the photographer. The positioning of the aerial shots comes from matching topo landmarks to the photos. This was an iterative process where I'd place ground objects and use them to place other photos and work out inconsistencies.
The ground shots can confirm object placement, but their highest value is in showing the height of trees and bushes. I might color-code those to make it clearer on the virtual terrain.
Once I have the images in place, I place trees initially in their approximate position and then adjust more finely as I flick photo visibility on and off. This is the same method astronomers use to detect celestial movement. I should have it done in another week or so.
I've put up a Youtube video showing the process.