[ODE] cylinder plane bug

Geoff Carlton gcarlton at iinet.net.au
Wed Sep 20 15:41:05 MST 2006


Hi,
I'm interested in these approaches, because they may lead to a whole new 
level of stabilisation.  Currently, bounce is mitigated by just slowing 
down the push apart forces and velocity.  A solution where bodies could 
push apart almost instantly yet not gain momentum would be a great step 
forward.

Unfortunately, I'm not really an expert on solvers so I'm not sure 
whether there are any fundamental drawbacks to Adam's approach.

Geoff


gl wrote:
>> Hmmm... this may be a controversial one:
>> Personally, I think that the current behaviour of not generating
>> contact points if the whole cylinder is behind the plane is the
>> desirable behaviour.
>>     
>
> I think that's wrong, the plane behaviour is important for eg. the 
> heightfield collider, as it avoids tunneling.  A better solution is an idea 
> I had a while ago, and which inspired Adam Moss to write an experimental 
> patch (that nobody showed much interest in).  My idea was to clear any 
> penetration correction velocities each step after they've been applied, so 
> once an object has been pushed out, it no longer goes flying into space and 
> beyond.  This would also mitigate the problem of 'spawning' objects in 
> penetrating states - they would just shuffle a bit instead of exploding.
>
> Adam's approach was slightly different, see:
> http://q12.org/pipermail/ode/2005-April/015705.html
> --
> gl 
>
> _______________________________________________
> ODE mailing list
> ODE at q12.org
> http://q12.org/mailman/listinfo/ode
>
>
>   


More information about the ODE mailing list