L3DT was mainly used to "inflate" a low-res version of the heightmap into a high-res detailed version (about 0.2m/px, in the final form), from which I pulled the Lighting and Shadow information to lay over my textures. Features like the roads, river, and building footprints were edited into the heighfield after the DM->HF generation. Then, trees and dropshadows were panted in.
http://postimage.org/image/ep5l8bolr/
