My review of the Nikon 70-200mm f/4G ED VR is in progress.
On the Nikon D800E, a focus difference of about one inch at a distance of 35 feet determines which lens is best. As I saw today, a “miss” of 1 inch “proves” that the same lens is better than itself or worse than itself. Such are the problems in making a comparison with a 36-megapixel camera. I am pondering the best way to handle this.
Contrast detect AF is provably not up to the accuracy required (as I have seen many times with various lenses); I can and did do better today using manual focus (my eye with a loupe, Live View magnified).
At ƒ/2.8, the 70-200/2.8 is too low contrast with possible focus shift to deal with, so it must be focused at ƒ/4. But at ƒ/4 as with the 70-200/4, there is too much depth of field for unambiguous focus. Focus bracketing is one solution, and I do have an 18-inch rail, so this might be the tedious and time consuming approach necessary, but it only works for distances not too far. Otherwise, it might have to be distance-only comparisons, with several focusing attempts. Given the lack of a tripod collar for the 70-200/4, this might have to be the proper way to go, regrettably ruling out mid-range shots, which are arguable more relevant.
All that said, this test image (bike) is revealing of performance nonetheless, and so I think I will present it as an aperture series for sharpness.
Unrelated: some new shots with the Nikon D800E + Nikon AF-S 24mm f/1.4G are posted on my cycling site. I liked using the 24/1.4G for product shots; I correct for distortion and chromatic aberration and the fast autofocus is helpful.