[hereiam2005]

Hard to tell. The diffraction of light is linked to the wavelength of the light. Since you never deal with monochromatic light, but resolution figures usually are given for monochromatic light, real world resolution will be lower than published data.

 

For e-line (wavelength of the light: 546.1nm, a green) diffraction limited resolution is:

f1.0 : 1501 lines/mm

f1.2 : 1251 lines/mm

f1.4 : 1072 lines/mm

f2 : 750 lines/mm

f2.8 : 536 lines/mm

f4 : 375 lines/mm

f5.6 : 268 lines/mm

f8: 188 lines/mm

f11: 136 lines/mm

f16 : 94 lines/mm

f22 : 68 lines/mm

(source: Nikon Ultra Micro Nikkor lenses brochure)

 

This is for diffraction limited lenses... So, lenses whose sharpness is limited by diffraction. In reality one does not find diffraction limited f2.8 or f2 lenses for DSLRs (maybe there are one or two exceptions, I do not know). Again, the numbers are for monochromatic light (e-line). Red has a longer wave length (around 700nm), and will cause resolution limited diffraction sooner (for f2 : 585 lines / mm, for f2.8 : 418 lines/mm (source:

 

For the different wave lengths for (monochromatic) blue, green and red you can find the optimal MP-count for different sized sensors here too: http://www.luminous-landscape.com/tutori...tion.shtml

If you take the green-yellow wavelength (550nm) they suggest as reference, they give for f2 on APS-C 209MP and for f2.8 on APS-C 107MP. But this still represents monochromatic light, and yellow and red wavelengths will lower diffraction limit more. 

 

Good post !

 

But then again, I used 700nm for diffraction limited calculation since I thought that we wanted something like a full color rendition, not green all over

 

Theoretical limit of diffraction (maximum resolution) of optical light is      (read more here http://en.wikipedia.org/wiki/Diffraction-limited_system)

 

the nsin(theta) is called Numerical Aperture in optical and antena engineering, or aperture in photography. In the old day the numerical aperture is around 1.4, and in good modern optic one can achieve NA < 1. Use f =1 and we get the diffraction limit.

 

I just take a number, for f1.0 the resolution is 1501 lines/mm, or d = 1/1501 ~  0.666 micron per line. One can see that this number is rather close to (and larger than) the half wavelength upper bound and the theoretical limit d of  0.5461/2 ~ 0.273 micron.

 

In theory, as long as we can increase the f stop, we can reduce d as much as we can. But this is not true in quantumn physics, where the lambda/2 is the hard wall ( due to something called the uncertainty principle, but I rather not go into detail here )

 

Nikon for once also makes industrial grade optical microscopes that goes for million $$$ apiece, so I am sure they are correct

 

Just my two cents btw.