I am having problems using point instancers along with semantic data, and I wanted to know what was the correct way of going about this. To give context, I have some rocks scattered inside a scene, and I am using a UsdGeom.PointInstancer for that. All of the rocks are being assigned a class and then added to the cache of the instancer. When I visualize the instance segmentation, the rocks that have the same model are considered as the same instance:
If I randomize their position, the scene looks like that:
This is weird seeing that the instance id segmentation does show them as individual objects.
Small detail there are 4 assets in the cache, so the times when I have more than 4 out off all the rocks are really puzzling me.
I am a bit confused as to what can cause the issue. Surely I am doing something wrong when I am creating/randomizing the instances. Could someone point me in the right direction?