I understand that every time a component is transformed there is a transform tree that can be multiplied out instead of calculating the transform per point. But it still needs to do these multiplications so what kind of accelerations can be made to compute the new location faster?
I understand that every time a component is transformed there is a transform tree that can be multiplied out instead of calculating the transform per point. But it still needs to do these multiplications so what kind of accelerations can be made to compute the new location faster?