[dirGames-L] 3D Snowboarding game

[Lucas Meijer]

Enrico Degrassi wrote:
> Rasmus,
>    we could do that indeed, but the physics engine is taking so much cpu 
> that when we exit from the closed loop, havok starts to give 
> unconsistent results.
> We need at least 14, 15 fps (the motion integration equation shoud be 
> done with timesteps under 100 milliseconts to give stable results)  to 
> get correct results from the physic engine and without closed loops this 
> seems very hard to obtain on slower machines.

That sounds like you could use some optimization in your havok setup.
As far as I could tell from playing the game, you would only need 1 
rigidbody for the player which is active, and all other rigidbodies can 
be pinned.

the havok xtra is lightning fast in situations where most rigidbodies 
are pinned, and only one or two are actually moving.

also, make sure to only use substeps=1 on the .step() call, using 
anything else is actually broken.

What kind of shape do you use for the player? is it convex, concave?

Bye, Lucas