[ODE] Collision CSG?

Anselm Hook anselm at hook.org
Thu Mar 27 11:04:01 2003


Another former MathEngine employee did a real time ray tracer using CSG
methods:

  http://home.earthlink.net/~howlers/bryan/quickray.html

I believe it does collision stuff but not things like penetration depth.

One nice use for a csg method would be for the pistons in a car engine...
a surface of potential face to face contacts with a high frequency...  By
intersecting the math instead of polys the labor could be more sane.

 - a

On Thu, 27 Mar 2003, gl wrote:

>
> I have been looking at a number of 3D models I've designed recently - very
> detailed large ones which may, for example, have players jumping and walking
> on them and which will move themselves - and it looks like getting good
> smooth collision response on some of them with primitives only will be hard.
> I'm sure it can be done by using lots of small prims to pad out the empty
> spaces left by the larges ones, but getting smooth transitions between them
> won't be that easy, and they may still not approximate the mesh shape well
> enough.
>
> I wondered for a while whether it's possible to do something akin to CSG
> with collision primitives - that would be really handy.  For example,
> imagine a hollow glass sphere that has small objects inside it.  If you
> could use a collision sphere, and then subtract a smaller collision sphere
> from it, you could do this with just two primitives.  I don't know if this
> is feasible though (contact generation might be hard) - any thoughts?
> --
> gl
>
> _______________________________________________
> ODE mailing list
> ODE@q12.org
> http://q12.org/mailman/listinfo/ode
>