I feel compelled to link to Michael Abrash's amazing and inspiring series of articles written when Quake was developed (don't know if there is a better link):

I feel compelled to link to Michael Abrash's amazing and inspiring series of articles written when Quake was developed (don't know if there is a better link): Ramblings in Realtime: http://www.bluesnews.com/abrash/ http://www.bluesnews.com/abrash/ I read this while I was in high-school and worked on my own 3D engine in parallel (inspired by Abrash's articles). As this was written while quake was under development, you get the reel feeling of the challenges they went through during development. Through "beam trees", z-buffer, to finally end at the PVS and sorted edge rasterization and mip-mapping technique that made the whole thing wrong smooth enough in the end. There is a similarly lengthy description of the considerations they had regarding shadow maps vs. e.g. gouraud shading. Carmack and Abrash's experience is very much trial-and-error and that's inspiring, because you realize that they went down many dead ends in order to find the right way to solve the problems. It's not some sort of divine stroke of genius, but far more their determination that solves the problems in the end. Of course Abrash also mentions some of the uglier hacks.