revolute joint bias

Please don't post Bullet support questions here, use the above forums instead.
Post Reply
Isawyou
Posts: 20
Joined: Wed Sep 02, 2015 10:00 pm

revolute joint bias

Post by Isawyou »

hi all!
what's the difference of revolute joint using the bias or the solve position constraint?
irlanrobson
Posts: 16
Joined: Sun Feb 08, 2015 3:21 pm
Location: Brazil

Re: revolute joint bias

Post by irlanrobson »

Basically: Stability. Using bias (e.g. Baumgarte) it is more easy for joints attached to bodies with large mass ratios to break or look spongy when the bodies are on top of other bodies and if angle limits are in, assuming a small number of velocity iterations. Problems with the spherical joint of the constraint aren't very noticible in most cases in my experience. But using a position solver (e.g. Full-NGS), convergence it is accelerated, and it gives good results in my experience. It also depends a little on how the constraints are solved, and the order. E.g solve multiple linear constraints together, direct enumeration, etc.

Here there is an example I wrote last year using PGS and baumgarte: https://github.com/irlanrobson/Bounce. It should be pretty straightforward to couple a position solver to this.
Post Reply