[ODE] Trimesh-Sphere inaccuracies

[ekt at libero.it]

i'm quite sure i've seen problems just standing still, without crossing edges, on trimesh made of big tris. i think they are two different issues.

the 'edge crossing jumping' lowers with smaller timestamps, but it's a pity having to double the physics steps just for this.

here is a proposal: i'll donate 100$ if adam or russ will take the time to look into this :) (well actually i'm going to donate anyway... it's just an excuse ;) ...and conitec of course could do much better than me ;)




> Thank you for the feedback. After some more hours of stepping through the 
> code it indeed appears to be a problem with edge crossings and not with the 
> size of the triangles. However, I have been unable to determine exactly what 
> goes wrong in these cases since the contacts don't look all that different 
> from regular ones. Yet the solver generates a larger correction force which 
> leads the wheels to bounce up a little more. Do this with a highly 
> tesselated mesh and you get a very bouncy car.
> 
> Regards,
>   Marco




____________________________________________________________
Navighi a 4 MEGA e i primi 3 mesi sono GRATIS. 
Scegli Libero Adsl Flat senza limiti su http://www.libero.it