I have rigidbody spheres with 3-axis rotation and low dynamic friction on the physic materials involved (~0.2). Angular velocities are in the 10-20 range (= magnitude of the angularvelocity vectors). If I use AddTorque to get the spheres rotating and they'll spin a bit, moving gradually faster in the direction of the rotation, and then suddenly they'll GRIP and sort of take off in that same direction. This is also evident when they're in the air: they can be falling straight down, but if there is rotation on a sphere before it hits the ground then the moment it hits the ground the grip will take hold and instantly send it along the ground in whatever direction it's rotating. Have I painted the picture? I hope so. This might all be correct physics behavior, though intuitively I want to see smoother, more laboured transitions. I just wondered if anyone had seen this in their own project and had found a way to smooth it out? I feels like a glitch to people who I've given my game to play.