Suppose I have a sphere and a cube intersecting with each other.

Is it efficient to store everything as two objects separately or should I run an algorithm to create the same visual result into a single geometric volume and store that instead?