From Terry Greer's page ( http://www.terrygreer.com/xcomapocalyse.html) "Every tile was designed to occupy a volume of space. Many tiles had multiple versions showing a succession of destroyed states. In the description off each tile there was a series of properties such as ‘health value’ and ‘destroyed tile type’. when the health value was eroded away the tile would then be replaced by the ‘destroyed tile type’ (which in turn could have its own health and destroyed type). Chains like this could be as long or short as required. Tiles could of course also bed invulnerable. This approach meant that potentially anything in the map could be blown apart and destroyed. This was made even more true in Apocalypse as tiles could also be flagged as supportive, and if a supportive tile was destroyed then those above it could also fall. Tiles were complex data structures. Each one also had a reference to line of sight definition – this was a simple 4x4x4 grid – a sort 3d texture – where each cell was either solid or empty. There were a relatively small number of these solid line of sight definitions, but enough to approximate the 3D shape of any shape of tile created. This meant that the world was effectively broken into a 3d voxel grid where weapon fire could be accurately ray-traced. It was a ball ache to set up initially, as every tile had to have a line of sight definition assigned, but really made life easy later on as collision in game then became automatically generated from the map editor. It was a genius approach, especially in the years before decent 3d raytracing – and one of the key identifying features of an xcom game. Players always loved being able to snipe enemy forces from right across the map through a couple of windows or blown open walls. It’s also a technique which I think useful to reinvent for use in conjunction with true 3D worlds and can think of lots of good mechanics you could use them for."