Are there certain applications that have conflicting goals, i.e. one use case is best suited by a certain representation, but another use case is suited by a different representation? How do we handle these tradeoffs, do we take a performance/functionality hit or convert between representations in some way?
In 3D, does our geometry become rectangle-based instead of triangle-based?
If you're dealing with multiple objects, is it feasible to use different representations for different objects? How can we then, say, intersect shapes using different representations?