[ODE] Not cylinders, but discs?

[Martijn Buijs]

Jon Watte (ODE) wrote:
> Cylinders aren't yet fully working. From what I understand, the main 
> reason to get cylinders is for car wheels that don't "stick out" on the 
> side of the car, with oil drum objects being a distant second.
> 
> Thinking about what you'll get out of car wheels, and how rays for 
> wheels work just fine, wouldn't it be feasible to do an infinitely thin 
> disc primitive? The hardest primitive to test against is usually 
> trimesh, but disc-triangle is just a line segment-circle in the plane of 
> the disc (assuming early rejection cases based on the disc plane have 
> all passed).
> 
> You'd model car wheels as a disc where the outer edge of your tires 
> would be; the inner edge, or even rolling surface, just aren't as 
> important for simulation.
> 
> I certainly don't have time to write this primitive, but is it something 
> to think about?
> 
> Cheers,

What's the reason the cone primitive has not been adopted? It seemed like a good alternative.

Random thought: would a sphere with two sliced off poles solve some of the computational issues that 
true cylinders have? It would also resemble tires more than a cylinder because it has a rounded 
contact surface.

Ellipsoids would also be interesting for this purposes.

Martijn