The sample I posted on PBase really astonished me. When I looked at the stuff on screen there was an incredible advantage to the larger file, but on the print it was nil. I showed the prints to a friend who was working at a pro lab and he could not tell them apart.
I never said that the thing Joseph Holmes have seen were not relevant to DSLRs. This has been indicated by Diglloyd quite a few times. When the OP decided to go for Nikon I pointed out the need to test everything. Nikon is known to having had problems with one of the AF-sensors being off on the D800. Thom Hogan has discussed this at length. Calibration is done with an 50/1.4 lens and process takes an hour as far as I recall.
As I said, I think the problem is less with DSLRs, simply because they are built to work with f/1.4 lenses. An f/1.4 lens needs twice the precision that f/2.8 lens has. As Diglloyd has pointed out, hanging a DSLR from your shoulders with a 300/2.8 is probably not beneficial to alignment of sensor and lens.
The other factor making Joseph Holmes observations important is that many buyers would use pre owned or rental equipment. If you read Joseph Holmes articles you will find that about two thirds of the samples he had seen had issues. The indication was that Phase backs coming from the factory were probably OK. Several of the lenses were bad, also Schneider and Rodenstock lenses coming from the factory.
Lens Rentals writes much about lenses. It seems that there is a lot of variation in lenses but really bad lenses are few. One interesting observation is that they test all lenses before shipping. They used to have manual inspection but since perhaps a year they are using Imatest. They have observed that lemons do pass trough visual inspection but show up in Imatest.
Once I tested mirror vibration. I made a series of shots with:
- Camera on tripod using Mirror Lock UP (MLU) and no AntiShake (AS)
- Camera on tripod no MLU no AS
- Camera on tripod no MLU and AS
- Camera hand held, no MLU and AS
These shots were done with a 200/2.8 zoom at medium apertures with exposures between 1/15 and 1/125.
To my surprise all the shots were sharp! The tripod/MLU shots were not sharper than free hand shots. Then I measured my test shots with Imatest. Imatest clearly showed that I lost half the sharpness on the non MLU shots. After seeing the numbers I went back to the screen and had a second look. After knowing the numerical difference I could see the difference on screen! I don't recall the camera I used, but I guess that it was a Sony Alpha 700 at 12 MP. Not using MLU made that camera into a 3 MP (half resolution squared) camera and I missed that at actual pixels. The experiment made me critical about relying on the eyes and religious about MLU.
difficult maybe... but not impossible for a trained eye, I think. In my experience the relation of image components to each other (color to color, gradation, sharpness, smoothness etc.) can be translated from real pixels viewed at 100% or 50% on screen to the respective print to some extend. Naturally a print looks different to a screen view... but if the processing chain is done right it's only the difference of 94 DPI (or whatever resolution of the screen) to 300 DPI (or 360 DPI or whatever resolution of the printer). Normally I can estimate how a capture (a digital file viewed on a screen that is) will look printed - at least this goes for my preferred printing paper in conjunction with the sharpening technique I prefer.
Anyway... a defocused MFD capture will always look like a super sized POS capture ... also in print. An accurately focused MFD capture makes the difference.
I do have the very same DB (P45) and when I saw Michael's capture made for that comparission I thought, okay, if this what you get from your P45, I'd prefer the G10. Because it's easier to use. But the capture is simply nowhere near of what a P45 is capable of.
With regard to the discussion in this thread... I've looked at some D800e RAW files available on the web and actually I am not impressed with regard to sharpness/details or so. I get that from my P45 all the time, even with my not so great lenses. I'd say the D800e is somehere in between my P21+ and my P45. But what really impresses me, however, is that you can push the D800e files at base ISO by 2, even 3 stops in the shadows without even having to think about applying any (luminance-) noise reduction. That is more or less impossible with my P45. Okay, I am comparing a 2005 DB to a 2012 DSLR. But still ... the D800e files I've seen look extremely good when pushed hard in post.
Then again... other than DR... they do not look nearly as nice (subjective perception !!! ) as the P45 captures with my lenses (Contax Zeiss, Digitars, Digarons). They look a bit more washed out (soft/dull) in image areas where I think the P45 would draw finer details/color differentiation. I am only talking about what I am seeing in the files. I really don't care whether or not a D800 or a Sony RX100 outperforms my P45 - I really don't care at all!
(and BTW... as a side note... I've also seen D800e captures suffering either from a tilted sensor or a misaligned lense ... so welcome to the world of high res imaging and warm greetings to J. Holmes ;-) ... ).