Still pressing on with the design bible, and I've started to dig into technical plans. At this point, i started to realize that my plans for multi-floored play really really doesn't work with the physics library I was planning for (Chipmunk Physics), so I'm currently torn between moving to Bullet Physics or canning multi-floored play. Going with Bullet would mean a lot more flexibility design-wise and less shoehorning. Going with Chipmunk would mean better performance, a more clean-cut game design, and giving Scott a good use-case for his library.
The only "real" case against Bullet is performance, and the worst-case scenario that I want to be able to drive is a 100-disc cloud with forceful interactions. That's actually likely to work with Bullet, since the official demo runs an 80-or-so body box stack. Running a "true" physics simulation means that I can take out a lot of edge cases from the collision logic in the level classes. In that case, level editors will have to provide a low-resolution collision and navigation mesh as before, but also have the freedom to go with a higher-poly rendering mesh. That's nice.
No comments:
Post a Comment