-snip-
I'm not the most knowledgeable in this department, so have patience with my ignorance, but doesn't a tile-based system allows for Toady to "cheat" more easily than a mesh-based one? Most physical discrepancies in DF are invisible, which is beneficial not merely in terms of immersion, but also in regards to optimization. Surely not having to calculate each object's mesh (even rather primitive ones, per your proposal) frees up a fair amount of processing power for DF?
It also means that terrain and structures can be mesh-based, which is better for simulating their destruction and can have their vertices used as pathing nodes (If a beeline to the target destination is not possible, the AI is definitely going to make a beeline towards a vertex).
Agreed, but I'm not sure that's viable for something with the scale of DF. Look at the FPS-damage done merely by flooding or by smoke. Imagine the processing-strain resulting from simulating all the minuscule bits scattered by a cave-in.
Yep,Fps would get bad fast with the scale of things in DF (especially after breaching hfs),im to tired to go into a bunch of details,so download Unity or UDK,and blender.Then add 100 plus low poly shapes (something with at least couple hundred triangles more then a cube[ A simple capsule shape has 960 traingles]) have them move around,then add your controls and try moving about,.That's not taking into account for fluid physics,terrain destruction,temperature,AI (which needs to be more robust to traverse 3D space then a simple grid as i recall),and all the fun stuff that now happens outside of the Fort.then there is the FPS issue,with 3d graphics low FPS can actually render the game impossible to play,or make the player feel sick,were in 2d graphics it makes the game feel more like a turn based game without making the player queasy