[ODE] Joint solving differences in Quickstep vs StepFast1

[Graham Fyffe]

Hi!  I'm not an expert on QuickStep but I have also noticed
differences, compared to WorldStep.  I have a detailed joint rig set
up for a car suspension, with several kinematic loops.  WorldStep
solves it beautifully at about 60 - 100 frames per second on my
machine.  QuickStep is entirely unable to handle the kinematic loops,
and the car just slumps to the ground.  Even setting the number of
iterations to 5,000 didn't solve the problem.  StepFast couldn't quite
handle it either.  So I ended up using WorldStep, which limits me to a
single car for the detailed joint rig.  But that's all I needed, so
yay!  :)

- Graham Fyffe




On Tue, 25 Jan 2005 12:30:48 +0500, Mohsin Hasan
<mohsin.hasan at trivor.com> wrote:
> I created a few prototypes of ragdolls and cloth simulation both using hinge
> joints. Both of the demos work fine with StepFast1 but if I try these demos
> with Quickstep the ragdolls start jittering (as if the joints are randomly
> being applied an impulse) and continue that way even after falling. The
> cloth on the other hand just explodes after a while. If I set ERP to a very
> low value (say 0.01) the jittering sort of gets very low and occasional but
> the results are not that good (mostly Ragdoll joints start breaking etc.).
> 
> Both of these samples work great with StepFast even with an ERP as high as
> 1.0. Any ideas what could be the issue of joint solving by QuickStep Vs
> StepFast? Anyone else also having same issues? I am willing to share the
> source code if anyone wishes.
> 
> Regards,
> Mohsin
> 
> _______________________________________________
> ODE mailing list
> ODE at q12.org
> http://q12.org/mailman/listinfo/ode
>