Let me first explain the test setup. The D700 was mounted on a
very sturdy tripod, mirror lock-up, and 10 second timer used for the
shutter release. The chart used was a 2' x 3' (roughly) ISO 12233
(2X). Lighting was full sun. The samples you see here have been
severely cropped to show 100% when clicked on to fully enlarge. The
reason I did this test was to find out how much detail I was losing
by stopping down to f16 and sometimes f22. The center focus point is
the circle inside the square. (See the original size below. It
covers a MUCH larger area) By shooting from this distance it is very
easy to see where the nyquist*
frequency is compromised by diffraction limiting and artifacts.
Chromatic aberration was optimized in Adobe Camera Raw 4.6. What is
really interesting to see is how moiré artifacts starts to be as
much of a problem as diffraction. My conclusion is that f11
seems to be the last stop where maximum sharpness is achieved.
However, please note that f16 is very useable and even f22 if it is
For an even better understanding of what
is happening, click on each image to enlarge it. Then "right click"
(PC) and "save as". In an editing program of your choice view at
200%. What is interesting is that at 100% the difference is hard to
see. This means that for a 12" x 18" print (or smaller) you simply
would be hard pressed to see a difference between f8 and f22.
A second question I have is how will a sensor that has the AA
filter removed perform (as some photographers have done)? My GUESS
is that moiré might well be a bigger problem than we see here - and
in fact then be the limiting factor?
If you are shooting landscapes moiré will probably be less of a
problem. In this case you might want to disregard the obvious moiré
patterns and concentrate on the resolution lines. Keep in mind,
though, that these artifacts are still there - and could show up
when least expected.
*Nyquist frequency is
defined as the highest spatial frequency where the CCD can still
faithfully record image detail. Beyond the Nyquist frequency,
aliasing occurs. Here is a full size JPG (level 10) of the f8
shot. Be careful, it's pretty big.
Phase II of this project will be to convert these raw files using a
variety of converters. I am most interested in how the various
converters handle the detail and moiré,