[ODE] small sized capsule on heightmap problem

rod rodkay at mullum.com.au
Fri Sep 29 04:54:26 MST 2006


On Thursday 28 September 2006 21:58, Martijn Buijs wrote:
>
> I did a variety of tests with very small objects before I submitted the
> code (things may have changed since) and never encountered similar
> problems. Those objects where similar in size as your capsule.

   I've played with test_heightfield a bit, and if I reduce the radius of the 
capsules, an instability shows up now and then ...


... from test_heightfield.cpp ...


		else if (cmd == 'c')
		{
			sides[0] *= 0.005;  // *** reduced from 0.5 !

			dMassSetCapsule (&m,DENSITY,3,sides[0],sides[1]);
			obj[i].geom[0] = dCreateCapsule (space,sides[0],sides[1]);
		}


  With the smaller radius, a capsule seems to fling off into space, or 
possibly disappear beneath the terrain.





>
> I suggest you check for negative geom dimensions (not sure if those are
> valid or handled well by ODE). Also, if the mass is to great the object
> would continuously penetrate, sink into the ground and "float" up again.

   Pretty sure the dimensions and masses are ok ... denisty of 1.0


>
> Have you tried replacing the terrain with a plane? dHeightfield uses planes
> for collisions internally, if it is a collision issue/bug it may occur with
> planes as well.

   Not yet ... might try if I get some spare time.


More information about the ODE mailing list