Former Raven colleague Matthew Breit and I re-united at LightBox Interactive in 2009. With restored co-worker status came the return of daily lunch discussions of all the side projects we wanted to do but would never find time to finish. One such project for Matt was building a Quake III level inspired by this piece by Peter Gric. Rather than tediously build the "disintegrating cubes" look by hand, he began developing a Python script that would generate cubes in Maya, then export back to Radiant. Matt's original plans called for stripping the walls off one of Bal's old Quake III DM maps and treating it as an art-only portfolio piece until I volunteered to design a new and fully-playable layout to go with it.
I began piecing together the layout in what little spare time I could find April 2010. Over the span of a few months we went back and forth on the design, settling on a monolithic concrete structure ripe for cubing. With the layout mostly done, Matt shifted his focus to finishing the CubeSpew script and began cubing the layout. That's when disaster struck. The level was so large and the cubes so dense that Quake III's renderer couldn't keep up - we'd either have to strip the layout down or build a new and smaller layout from scratch.
Rather than gut an otherwise strong level, we decided to set the original layout aside for later use and I started work on a new, slimmer design. Much like Q4DM10, I began with the idea of a central jump pad flinging players from one end of the level to the other, this time arcing through a hole in a massive chunk of concrete towering over the level. The new layout also retains a three-tiered structure from the original, this time relying more on short jump pads instead of long, size-bloating, and difficult to cube ramps. The new size played excellently in a 3-player playtest, and Matt started cubing once more.
In final form, the level still suffers from some low performance spikes due to the density of cubes (Quake III's renderer doesn't handle "throw a bajillion triangles at each core" batching the way new video cards prefer). Still, by way of Matt's cubes, textures, and lighting, this is without a doubt the most visually striking level I've ever worked on and it also stands as one of my stronger pure DM layouts.
Matt released an updated version of this level on March 19th, 2012 in which he updated the CubeSpew script to generate variable-resolution lightmap scaling on faces (for better-defined shadows) and hooked up an AAS file for basic bot navigation. While the standard Quake III bots don't handle the numerous ledges and jumps well, it does at least make the level playable offline.
For more on this level, including technical cube details, visit lunaran.com
I built this level as part of MapCore's 20 Brush Challenge in which we had two weeks to build a level using only 20 solids (skybox and game entities such as triggers did not count toward this limit).
I blocked out a few rough passes before deciding to take a chance on the interlocking hexagon style which gave me exactly twenty platforms with which to assemble my layout. I used a mix of jump pads and func_bobbing platforms to ensure connectivity between platform clusters and different elevations, and ended up with a fairly solid loop through the level.
I had originally blocked out the level using Matthew Breit's lun3dm5 concrete textures, which were of course reserved for dm5's eventual cubing and release. Matthew was kind enough to give me a dirtier concrete set with a fancy jump pad texture, while also running the skybox through a Python script he developed for lun3dm5 to generate the sky lighting. His final contribution to the level was suggesting I take the name "Hexas" to its logical conclusion.
Don't Mess With Hexas received a handful of votes across the different categories, but sadly came up empty in the final results.