autonomousvision/sdfstudio

get_surface_sliding vs get_surface_occupancy

iszihan opened this issue · 1 comments

Just curious what's the main difference between get_surface_sliding and get_surface_occupancy? I know that the latter one uses a level set value of 0.5 because it is extracting a mesh from an occupancy network. But can I also just use get_surface_sliding with a level set value of 0.5 for occupancy based network like unisurf?

Hi, get_surface_sliding use multi-scale strategy to skip empty space since if sdf is very large, it means a large region centred at this point is empty. However, occupancy only have the inside outside information, the multi-scale strategy might not work well.