Yeah, where the concept is “simplistic” in real world, the translation to fictive is a major challenge as you continuously need to calculate distances and keep it seamless. Allthough the link with varifocus can be made. I should have added distortion and dynamically chosen object removal (focus culling) 
True, this is also where your mentioned metering modes come into play.
I can definitely imagine, it’s not just a simple one factor solution.
Edit: (I’m derailing, appologies…)