As an indie game you likely have to take the sparse route, just enough to be suggestive, but picking what exactly is suggestive enough, and also believable, is a lot of tough decisions, even more so in 3D, which is what you're apparently working with here.
Right. And I'm no 3D modeler, but I've gotten away with Sketchup (for modeling) and LithUnwrap (for UV editing). And the theme is super barebones anyway (like, seriously, Tron-style edge glow). It's all supposed to be a VR representation of underlying hardware; utilitarian. But the potential is there for allowing the player to own "themes" that would replace all the bare-bones models with some kind of "sculpting" that would make it look like whatever. It'd be trivially easy to add, codewise, but a massive undertaking 3D model and texture wise (resources are loaded via string constant, so having a directory that's supplied via runtime variable is trivial).
Of course, with 3D the approach can also be quite different than with something like Cogmind. Here I mostly just come up with an ASCII pattern and give it a name that fits the theme, whereas in your case you have the option of just modeling/drawing something and throwing it in there without even a name (as long as it has no bearing on the mechanics or other content).
Certainly something I'll have to do. I'm just not good at coming up with random doodly dads without having some idea of what I'm trying to actually
make. Some of the themed constructs would be a lot easier to just insert-random-thing-here, but I'm starting with the utilitarian view first. The "themes" shouldn't really give additional cover or obstacles, just differentize them.
Anyway, thanks for the input, even if it was "you just make stuff up."
I'm super proud of those maze layouts too, even if it's really simple process. Just the Mazectric rules for Conway's game of life (B3/S1234) run about 250 generations with a little pruning along the outer edges (doors auto-set the cell to "dead" so that they remain passable, floodfill from the first door, then disconnected areas are discovered and connected via best-guess) which allows the banks to line the walls or create dead ends, yet still have every machine be reachable (the ones in a corer
are reachable, diagonally). In the rare case that an open (dead) cell is truely unreachable via best-guess connection, it's left alone and the global pathfinding map has it marked as filled (same as the cell a machine occupies) just so the AIs don't try to path there (not that it's a problem yet as no AI code has been written to perform NPC duties beyond simple patrol routes).
Oh, and the wide hallways are the standard routes, the narrow paths are the secret passages we all know and love. Technically they serve the reverse purpose as Cogmind's secret doors, letting the player sneak about in ways that the guards cannot follow, but fill the same game-mechanics space as the air vents so commonly used in stealth games. Its also hard to tell, but they'll go up-and-over the regular passages and will connect between floors (screenshot shows only a single floor complex, but multiple floors are possible).
And due to the way they generate, it is possible for me to dictate how many of them spawn, and eventually subselect from this (representing backdoors located and removed by the complex) or add new ones. Removing one secret path might actually
mutate a different one (as they generate iteritively and try to avoid one another). Which is actually really clever from a theming point of view. You can see one path that tried to take a short dogleg towards its desired destination, found a room in the way, and went around to the right. If the total path length so far had been long enough, it would have terminated prematurely on one of those two rooms.
...I should stop rambling, I think I'm maybe a little too excited about what I'm doing and have no one to share it with. XD
