[ODE] Jitters
Geoff Carlton
gcarlton at iinet.net.au
Tue Nov 9 12:44:00 MST 2004
Be careful of huge mass differences and in particular very small
masses. I guess you can't help it if you really do have a huge
difference (e.g. a church tumbling around in the air). However you
could group objects into small/huge sets and special case the result so
that a "basketball-church" collision is treated as "basketball-null".
Thats probably going to work better than any mass difference could do,
and it means you can keep the masses around 1-100 units (which is what I
use and find to be stable).
Tim Rightnour wrote:
>On 08-Nov-2004 Geoff Carlton wrote:
>
>
>>What is the mass? If its very small then even 0.01x the velocity can be
>>too great a force (I've seen this with very small objects). Try a lower
>>value for both linear and angular, and see if it fixes it, (e.g. -0.001).
>>
>>Judging by the results it looks like the -0.01 is providing way too much
>>force and the object is flying in the opposite direction each frame.
>>
>>
>
>Looks like you were dead on. I guess I forgot that I was applying force
>directly to the object, not just modifying it's linear velocity directly. My
>masses are all rather minute, due to me trying to keep them between 0 and 10.
>It means my light objects all have infitesimally small masses because they got
>normalized against things like houses and churches.
>
>I still can't make the jitters go away, but perhaps if I play with the damping
>forces some more, I can find a value that magically works.
>
>Stupid me. My old physics code just had me working directly with linear
>velocity, not applying force, I guess I got it confused in my head. Thanks to
>everyone for helping me find where I went wrong with the damping.
>
>---
>Tim Rightnour <root at garbled.net>
>NetBSD: Free multi-architecture OS http://www.netbsd.org/
>NetBSD supported hardware database: http://mail-index.netbsd.org/cgi-bin/hw.cgi
>_______________________________________________
>ODE mailing list
>ODE at q12.org
>http://q12.org/mailman/listinfo/ode
>
>
>
>
More information about the ODE
mailing list