zSolver1Shape Error

Sitingliu

Hey, can anyone help on this
When we do simulation of the Fascia layer,
at certain frame, the fascia just explode to really far distance.
and an Error message shows, ERR| Could not compute zSolver 'zSolver1Shape'. Solver has encountered instabilities.

Does anyone know what's causing this issue

Thanks

Siting Liu

Sitingliu

and additional Error message as
ERR| Error: xSYEV solver returned non-zero exist code: 8. //
ERR| Error: xSYEV solver returned non-zero exist code: 8. //
ERR| Error: xSYEV solver returned non-zero exist code: 8. //
..................
and
ERR| Matrix size is 9 x 9 //
......

and
ERR| The problem input matrix
A = [nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, ]
..............................
..............................
..............................

Essex

It's difficult to diagnose this without being able to reproduce it locally. The error message likely means that the position of a vertex is NaN (not a number). That likely means that something failed some time earlier. Likely causes for this are triangles becoming degenerate and then something dividing by zero somewhere.

Today, the thing to do is look closely at what's happening on the frame before this happens and try to find things that look bad. Perhaps try narrowing it down by turning off collisions for just that frame, or taking smaller timesteps, or looking part way through the frame. Try out the mesh-analysis tool on your bone shapes on that frame. What sort of stuff have you already tried?

Moving forward, we continue to improve these sorts of error detection issues. Every time this sort of thing comes up and I'm able to determine what went wrong, I've been eliminating the problem or adding a check to find it earlier and print a better description. The next version might provide a more helpful message.

Sitingliu

Hey Essex
Thanks a lot for the response, I have figured out it was the animation controller gimbal between two key frame next to each other (example: frame 110, frame 111), which I couldn't notice it from the playblast. The body parts actually twist like crazy inbetween the two frames, which might be the reason to cause this issue with my zSolver substep 2.
I guess this might be the reason.
After I use the Euler filters in the graphic editor for the animation curve, and fix the controller twist a bit. Everything is fine now.
I really feel like after all the technical tasks went through, there is still a bit challenge ahead of this muscle simulation, is to collaborate with animator animate it in a anatomical way.

Yes, more detailed Error message feedback from the next Ziva update will be fantastic

Siting Liu