[ODE] Particle/cloth/skeleton (was: Re: Choleski factorization)

[Martin C. Martin]

Antonio_Martini@scee.net wrote:
> 
> modeling rigid bodies as particles is rather unconvenient as they generate
> really many constraints which in turn means that really many iterations are
> needed
> in order to achieve an acceptable error(objects changing volume with
> highspeed impacts, etc..)
> 
> Not to mention problems in the collision response. Talking about rigid
> bodies i would steer
> well away from that method.

Except that each constraint needs only 4 floating point multiplies and 1
division to enforce.  Which means you can simulate many, many iterations
quickly.  Also, ball-and-socket joints as well as hinge joints actually
reduce the number of particles with either reducing (hinge) or not
changing (ball-and-socket) the number of constraints.  In practice, a
dozen iterations seems to be more than enough, at least for simple
systems.  It can also be made adaptive, iterating until the error drops
below some specified value.

- Martin