[ODE] Newbie Question on Numerical Integration Method

Alen Ladavac alenl-ml at croteam.com
Tue Jun 7 08:37:24 MST 2005

Hi Piotr,

Joint feedback forces are wrong because position correction in constraints
is mixed with velocity correction, giveing you larger forces. You might want
to look into some recently mentioned modifications (see threads "Patch for
eliminating penetration bounce mega-force?" and "patch - position-correcting
constraints", though you might need to dig some more backwards in time).
Note that I didn't actually try this myself, so perhaps I cannot be sure
that it is _the_ complete solution, as there might be some other reasons for
wrong forces that pop up later, but this should at least help to get more
correct forces, regardless of the solver used.


----- Original Message -----
From: "Piotr Obrzut" <piotr_obrzut at o2.pl>
To: <ode at q12.org>
Sent: Monday, June 06, 2005 19:11
Subject: Re[2]: [ODE] Newbie Question on Numerical Integration Method

> Hi,
> 6 czerwca 2005 (19:10:54) Jon wrote:
> JWO> The ODE solver uses a first-order forward integrator; it cannot
> JWO> methods (as doing the LCP solver to a higher order would be
> JWO> significantly harder...)
> JWO> If you have "wiggle" on resting bodies, the first thing to look at is
> JWO> your time step. You should be using a fixed time step, and step
> JWO> times to catch up to real time, but no more. The reason for this is
> JWO> bodies "resting" in ODE really are continually falling into the
> JWO> and getting a penalty force applied that pushes them out. If the time
> JWO> step varies, they fall different lengths each timestep, and get
> JWO> different forces applied, which ends up causing instability.
> Right I also need very stable simulation and I'm using min and max
> steps count and when min == max then simulation is very stable, when
> min << max then it tends to explode. Still I have some problems
> with joint feedback forces: when min << max then they are very large,
> when min==max they are much lower but still they sometimes go to
> the large values. For me it is importand to have joint feedback forces
> on constant low level - anyone could help me with that?
> --
> greetings,
>  Piotr Obrzut
> _______________________________________________
> ODE mailing list
> ODE at q12.org
> http://q12.org/mailman/listinfo/ode

More information about the ODE mailing list