[Bf-committers] ODE

[bf-committers@blender.org]

On Fri, Dec 20, 2002 at 12:14:48AM +0100, Carsten Wartmann wrote:
> nlin@nlin.net writes:
> 
> Norman! Nice to hear from you. I wrote an email to you some days ago.

Danke fuer die Gruesse :-) I'll respond to your message ASAP.

> Oh? I was sure to see a difference when I set the Bowling pins to Box,
> Cylinder and Polyhedra.

Hm, I am fairly certain that Polyhedra cannot be working at all, since
the triangle-collision (i.e. arbitrary polyhedron) API is even now still
undergoing rapid evolution in ODE. What I suspect might have happened is
that your different choices caused the center of mass to be displaced,
causing different behavior, but still using a box or a sphere for collision.
There really needs to be a debug mode in the game engine where you can see
the physics collision objects being used (this would also help to debug
the problem with incorrect radii being used when generating the sphere).

>  > interest is the item commenting on the slowness when lots of static geometry
>  > is present. This should theoretically be fixed in the current version of
>  > ODE, which uses hierarchical collision detection.
> 
> Ah thats fine. What does "theoretically" mean? ;-)

This means that the ODE API changed to improve its collision detection around
November 2002, and that according to the ODE docs, this improves performance
considerably. However I have had no time to test this assertion in an isolated
test program, let alone even think about integrating this into Blender.
Volunteers? :-)

-Norman