Warping my reference mesh to validate model

1

Hello shapeworks community,

I am developing an SSM of vertebrea and have been following the deepssm example for python.
After fitting and optimization I reconstruct all the meshes from the reference.
In most cases, it does a great job.
However, there are some cases where there are connections between particles that are not desired (example attached: the particles within the red circles should not be connected).
I was wondering:

  1. what is the best way to avoid these errors? Should I further groom the meshes before fitting the SSM?
  2. Is there a metric to spot these errors automatically?

Thank you very much,
Best wishes,
Simão

Screenshot 2023-09-27 at 19.02.08
Screenshot 2023-09-27 at 19.02.081920×1302 59.6 KB

2

Usually this type of bad warping is due to the particle system not being in good correspondence. I would check that first. When you look at the PCA modes, are the particles crossing at all or do they stay in correspondence across the PCA values?