If I recall correctly, in WC2, the entities were modeled in clay/plasticine, photographed from several (8?) distinct fixed angles, and then artists hand-painted cels over the photos. The cels were then scanned and digitized into 2d sprites. The maps were flat images with hidden metadata layers to define land and water boundaries.

I also recall that you could replace the peons' wood-chopping sounds with fart noises, which was hilarious for almost 30 whole minutes.