Verlet and Levels of Detail
-
- Posts: 1
- Joined: Wed Nov 21, 2007 10:26 pm
Verlet and Levels of Detail
I am working on a RTS style minigame that features top down command of a variety of intelligent vehicles and players. The details of implementing the actual vehicles and players are not my responsibility but the work of the collision detection and physics are mine. The terrain will be a trimesh or heightmap of some sort that represents complex mountain ranges and or dunes in the desert. The modeling of destructible terrain is optional but desired. The main issue that I am faced with is this: most physics engines are designed around fairly small simulation areas. My simulation area can cover an area of up to a thousand square kilometers. I have the idea of using a verlet based integrator to do my physics computation due to the speed of calculation.I would like the ability to change the level of detail of my models on the fly, changing out a simple box with a single wheel moving across a plane to a fully articulated truck with springs and tires. For the higher level of detail, I may need to replace the verlet integrator with something a little more capable. My primary question is that of collision detection. Given that my vehicles will be constructed of convex trimeshes, what is the best way to handle the collisions between the ground plane and the actors?