I had an idea, we can slide the axis say from left to right, when find two primitives' shape is so overlapped that one of them reaches centroid, we split once and keep moving. This way we can ensure primitives that are close to each other are in one bounding box
spookyspider
So in this method, we are partitioning by primitive (or by primitives), rather than by space? If you have overlapping primitives then, you might have to make sure they're in the same partition, or maybe that doesn't matter so much?
I had an idea, we can slide the axis say from left to right, when find two primitives' shape is so overlapped that one of them reaches centroid, we split once and keep moving. This way we can ensure primitives that are close to each other are in one bounding box
So in this method, we are partitioning by primitive (or by primitives), rather than by space? If you have overlapping primitives then, you might have to make sure they're in the same partition, or maybe that doesn't matter so much?