Plastic deformation time-step sensitivity.

[digbeta]

Hi, Miles and Friends,

I've noticed an issue in FleX whereby plastic deformations don't appear to occur in certain circumstances depending on the value for the dt time step passed to NvFlexUpdateSolver.

If, for example, I pass the actual elapsed time (g_realdt in your example demo) to NvFlexUpdateSolver in my application (which is simply a falling body), a softbody with plastic values appears stiff, showing no deformation. However, if I instead pass a value of 3 X the elapsed time, with all other values remaining the same, the body crushes/deforms. I understand some of the other guidance regarding the impact that increasing substeps has on making constraints appear more rigid.

Any ideas you can share of what might be happening and if it can be resolved?

Thank you!