Clarification for Coarse-to-fine train step

[SamyakR99]

Hi Gengshan,

I was looking at the paper and in that it was mentioned that the step S0 does not have any bones and we start with a sphere.

When I looked at the template.sh file

here it looks like the bones are initialized to 21 (B=20). I am slightly confused if we call this the step S0 because at the start we should set it to 1 (B=0). (A general trend I observed was n_faces did not align with the n_bones)

I am not sure if I am missing something here. I would really appreciate if you could help with this. Thank you!

[gengshan-y]

Hi Samyak, sorry, the issue slipped through.

The results in the figure were produced with B=0 in the first stage (and corresponding B and M for the rest of the stages). And I modified those hyperparameters when releasing the code to make optimization more robust.

More bones at the beginning make optimization more robust to large deformations, but more susceptible to wrong root rotation estimations.

The difference in the number of faces should not cause a big difference in practice, as long as it is increasing.