Hey Craig,
Sounds great and it shows that the scene was too large for the wanted DOF. There are tons of DOF calculators available. I have one on my Mac (Lens•Lab) which is also available for portables. Anyway, the idea was a while ago to keep things in scale (before that units where just an idea, nothing real). But the more we go to the simulation of photography/film, the more I suggest to think about that.
The film gate of the largest film camera is to my knowledge below 3600 ;o) http://robroy.dyndns.info/lawrence/mammoth.html
Anyway, your question about the start and end, this is normally needed to describe the start and the end point of the black and white values of the Depth-Pass in Multi-Pass rendering. It has nothing to do (AFAIK) with the Physical correct Render, other wise it wouldn’t be physical correct, as we can have it not both ways. ;o)
In the render settings we have the options for that gradient based result to change the progression of the DOF in the standard renderer. Pure simulation and not really close to the physical render.
The physical render has the huge advantage to calculate motion blur and DOF at the same time, and as I have demonstrated some years ago, it does calculate based on the film gate, which means, looks around “corners” so to say, in opposite to a pinhole camera simulation from the earlier days. Motion Blur AND DOF is nearly impossible to achieve in “post” later on, as we miss data and secondly once we do one pass the other one is nearly impossible. (There are some tutorials available that suggest other wise. Well, compare the results, so I do not have to judge this stuff. ;o)
BTW, I’m not bothered by anything that will not work. I keep it with NLP, there are no failures, there is only feedback. :o)
Have a great day
Sassi