Real-Time Physics Simulation Forum

Re: Resting Contact: Relative Velocity < Impulse Threshold

2013-03-03T21:23:45+00:00

I just ripped off Henge3D.

Statistics: Posted by SinisterMephisto — Sun Mar 03, 2013 9:23 pm

Re: Resting Contact: Relative Velocity < Impulse Threshold

2013-03-03T08:54:12+00:00

So far i have managed to get the collision response (bounce) to match physx
but still lost on numbers to used for resting contact.

i current use value of 0.8f to 5.7f of K and s 0.001f - 3.65f

Statistics: Posted by SinisterMephisto — Sun Mar 03, 2013 8:54 am

Re: Resting Contact: Relative Velocity < Impulse Threshold

2013-02-28T00:37:53+00:00

They're just tuning values. They don't necessarily correspond to any real values for things (the damping and spring constants for resting contacts have real world analogies, except that resting contacts aren't really springs, so again the values are just for tuning). Play with the values until you get behavior you like.

Statistics: Posted by Numsgil — Thu Feb 28, 2013 12:37 am

Re: Resting Contact: Relative Velocity < Impulse Threshold

2013-02-19T14:20:12+00:00

PS if there is a different way to do this please enlighten me.
It seems this method is based on a paper by Baraff

Statistics: Posted by SinisterMephisto — Tue Feb 19, 2013 2:20 pm

Resting Contact: Relative Velocity < Impulse Threshold

2013-02-19T00:35:13+00:00

Hi guys
I was building a re-targeting engine following Chris Hecker's Spore paper then it occurred to me i could basically use the same Verlet system for my ragdoll.
I decided to follow Rick Baltman's GDC talk slide and paper to implement this.
http://www.docstoc.com/docs/37750453/ar ... 2004baltma
http://www.docstoc.com/docs/40467261/Ve ... -Degree-of

In his paper just like in Bullet there is this magic variable called Impulse Threshold.
When relative Impact speed < Threshold = rest
When relative Impact speed > Threshold = bounce
Any idea what this value is?
I used a value 0.025 in Unity3d and it was the closest i could match the built-in physX engine.

I scoured the box2d sdk and could not find a value for this either

That aside, In his paper he uses a spring system for resting contacts.

Code:

Vector3 Force_penetration = (K_Spring * penetration)  - (K_Damping * penetration_);

The spring and damper variables are equally a mystery to me.

Statistics: Posted by SinisterMephisto — Tue Feb 19, 2013 12:35 am