Originally Posted by BillDrew
To pick a nit: diffraction effects not changing with focal length might be reasonable for a specified sensor/film size, but not otherwise. The old timers who used 8x10" (or larger) format cameras shot landscapes at f/64. If you tried that with an APS sized sensor the results would not be good - as shown at the site Mark linked to.
Quite true ...
There are two independent issues here. One is the resolution/diffraction limit set by the sensor. This is controlled primarily by the photosite size (not
the megapixel size as stated in the post that gives a link to a good description where the concluding chart clearly shows that the D3 and D300, both 12mp cameras, have radically different critical f/stops). This limit is independent of the focal length.
There is a second diffraction limitation controlled purely by the lens' characteristics. Unlike the sensor-based limitation, this diffraction limitation issue, know properly as Dawes' Limit, is a factor of the actual physical size of the aperture relative to the frequency of the light being focused. This translates to the limit occuring at different f/stops for lenses of differing focal lengths.
The bottom lines is that to get the sharpest results, you should not stop down below either limit. With short focal length lenses, the lens' limit may occur at a wider aperture than the sensor's and stopping down as far as the sensor's limit would not produce the sharpest results at the point of best focus.
BTW, SLR Gear's review of the 18-200 Nikkor VR (they haven't tested the new VR-II version yet) indicates optimum performance at between f/4-f/5.6 at the wide end and around f/8 at middle and longer focal lengths. Stopping down further begins to significantly reduce sharpness as diffraction takes its toll.