Don't agree with that - sorry.
DoF is a rendering issue, not image capture. A lens is only ever focused at a single distance. Anything closer or further away is more or less out of focus. It 's only when you render that it becomes visible if it's in or out of focus.
For me, pixel peeping at 1920 pixels on a 24" monitor which is 55cm across at 50cm away this effectively means I'm looking at a print which is 150cm x 100cm at 50 cm. This is way larger than used for DoF calculations.
From:
http://en.wikipedia.org/wiki/Acceptable_circle_of_confusion#Circle_of_confusion_diameter_limit_in_photography
you can see the CoC is approximately given by:
CoC (mm) = viewing distance (cm) / desired final-image resolution (lp/mm) for a 25 cm viewing distance / enlargement / 25
A 5D2 sensor is 3.6cm on the long side, so the magnification is 150/3.6 ~ 42. Assuming I'm going for the "standard" 5 lp/mm @ 25cm and a viewing distance of 50cm, this makes my CoC
CoC = 50/ 5 / 42 / 25 ~ 0.0095mm.
The "standard" CoC for a full frame camera is ~0.03mm so I've a CoC which is more than 3 x smaller. Looking at DoF Master, the nearest camera to this CoC is the Powershot Pro1 which has CoC of 0.008mm.
Dropping this camera into DoF master (
http://dofmaster.com/dofjs.html) and assuming a subject distance of 4m (not sure what the actual focal distance was but...) at 17mm and F4 gives:
Subject distance 4 m
Depth of field
Near limit 2.78 m
Far limit 7.16 m
Total 4.38 m
So, depending on how big you look at the image and from how far away will determine how much DoF you perceive. At my viewing size and my viewing distance, I'd need to drop to F/9.5 before I'd get a DoF that appeared to be infinite. In other words, I'd expect to be able to see things in the distance out of focus.
Looking at the image again with that knowledge, it's actually a pretty good lens (from memory, the far part of the ceiling would have been ~15m away...).
