Here is a simple reproducible script computing polygroups from Angle and then setting materials for each poly group (purely for visualization).
Running this script on a simple cylinder results in only one side correctly being computed.
I can use the same exact angle filtering for SplitNormals and get perfect results, but compute poly groups by angle is bugged.
On more complicated geometries it fails worse!
A hacky work-around is to set Min Group Size to 1 which (following the source code) disables to the PostProcessing function in the FPolygroupsGenerator. This function seems to be squashing multiple polygroups on to each other smh.
Thanks, setting the group size to one is working. Took me hours to finds out whats wrong
