On closer inspection, though, are you perhaps referring to superimposing foreground glyphs across multiple layers? (The engine is optimized on the premise that only one glyph and one hue can be used per cell.)
Correct. Think about power lines laid out on the ground. If a robot standing on them isn't totally opaque, it'd be nice if the power lines were visible beneath it.
Hey Decker, didn't know you were using REXPaint 
. What kind of UI design are you doing with it?
This is kind of embarrassing. For several years, I worked on a little roguelike (I made some prototypes with OpenGL) that I ultimately abandoned because my ideas (mechanics) were not good enough.
One of the thing I enjoy most about games (probably even more than playing them!) is figuring out optimal builds. That was the main focus of my game: figuring out builds to be able to survive each level, in a rich tactical environment.
Cogmind reminds me of the game I was developing, so I have a particular affinity for it.
Recently, I thought about it some more and I figured out where I went wrong.
1) 3D grid representation with heightmap -> too much work.
2) Complex mechanics are bad. Emergent complexity, not inherent complexity.
3) UI first. Not thinking about the UI experience leads to bad design. It drives everything.
4) It shouldn't be a roguelike. Controlling a party allows richer tactical interactions.
5) The story doesn't matter all that much.
6) Focus on one thing and do it right. Examples: Cogmind, FTL, Portal, Wesnoth, Defense Grid, TOME.
The party-based setup really brings out the tactical depth that I wanted but couldn't achieve with a single character. The core elements include tower defense, stealth, player-designed turrets, robots and biological horrors, and simple but rich powers: polymorph, teleport, tow, transport, mine, infect, mind-control, hack, regenerate, etc.
For a long time I didn't want to have a party, because it's a chore to move several creatures over a long distance. I would lead to a game like Wesnoth where your units are mostly limited to a small area so it's not too annoying to manage. I wanted more flexibility.
I made a radical design change to achieve this: control as a resource. You can control about 2-3 units per turn from the central AI (but most units are autonomous/immobile). As in Cogmind, time is working against you. It leads to agonizing decisions (move a unit out of harm's way, build, attack, bring in reinforcements?). You can tow/carry units and send GOTO orders to optimize time, and use power-ups to temporarily gain more control. All-in-all, I think that design could achieve both high emergent complexity and a low level of micromanagement. A bit like chess, but a lot more fluid.
I have no plan to code that game at short to middle term. I'm just documenting things every few months when I get an epiphany. I used Rexpaint to figure out some UI measurements and avoid some resolution issues that you had
My priorities right now are that volley simulator, and neural network/AGI development.
And that's a longer post than I had anticipated
