In Physics.OverlapSphere documentation there is a note that says:
NOTE: Currently this only checks against the bounding volumes of the colliders not against the actual colliders
Can anyone explain the difference between the two?
My understanding of that phrase is that OverlapSphere will not perform an accurate test for whether mesh colliders lie within the sphere, say, but instead it will test against a simpler convex bounding volume collider that fully covers the object.
If your objects already use box, capsule, or sphere colliders, you should notice no difference, as these are already bounding volumes. What’s unclear is what bounding volume OverlapSphere will use in place of other more complex colliders - I would guess they’d be substituted for a box collider, but sphere or capsule would be just as valid from that definition.