[Yamanobori]

That article is not about the diffraction limit. The diffraction limit is related to the permissible circle of confusion which is determined by format size.

[MetaD]

The Luminous Landscape article mentioned by Westmill pointed to this Cambridge in Colour article, which has a graphic that lets you change aperture and pixel size, thereby viewing the how many pixels are affected by diffraction.

Just roll over the two factors, aperture or camera (pixel-size).

[Yamanobori]

MetaD, and if you scroll down there is a diffraction limit calculator which calculates the limit based on format size, not pixel pitch.

[MetaD]

Yamanobori, Yes, there is a second calculator on the Cambridge Colour page.

By changing the inputs, it shows that the "airy disk" (a direct measure of diffraction) depends on aperture, but not at all on format size, which is what I said up above. The calculator is cool because (in the advanced mode) you can plug in camera parameters, like 35mm Full Frame with 36 mpixels. This gives you the pixel size, and then you can easily see whether the airy disk (diffraction circle ) is larger or smaller than the pixel size.

The Calculator also calculates Circle of Confusion which depends on format and pixel size, but not on aperture. It is related to depth of field, but CoC and DoF are perceptual rather than mathematical measures. Cambridge in Colour defines Diffraction Limit as the point where airy disk exceeds CoC as defined by their calculator.

Presenting the diameter of the CoC and the Airy disk in microns let's you understand how those two effects appear at a given camera's pixel size. By playing around with the calculator, you can see that a small sensor camera has greater depth of field than a medium format camera.

Another take-away is that CoC is defined a bit too coarsely for the latest FF small pixel cameras like the D800. With such high resolution, you should set Depth of Field narrower than the aperture markings on the lens based on the older definition of CoC (whatever, something like 20/20 vision at 20 inches from a 20 inch print?)

If Yamanobori and I have any differences, I think it is about definitions, not the math/science.

The Diffraction Limit is different from noticing that diffraction has started to reduce contrast on fine details in your image. I highly recommend getting a subscription to Diglloyd's "Making Sharp Images". Today's 36mpix FF cameras have such high resolution and small pixel sizes that they compete with Medium Format, but only if you have superlative lenses and superlative technique. It is possible that Nikon has accidentally produced a new camera that requires people to buy Zeiss or Leica quality lenses in order to fully utilize 36mpix.

[Yamanobori]

Another take-away is that CoC is defined a bit too coarsely for the latest FF small pixel cameras like the D800. With such high resolution, you should set Depth of Field narrower than the aperture markings on the lens based on the older definition of CoC (whatever, something like 20/20 vision at 20 inches from a 20 inch print?)

Actually, the difference is about the definition and by extension the math and science, which is pretty clear. I don't think you understand the basic concept of permissible circles of confusion. The pixel size will have no impact on image sharpness. Neither will it have any effect on the diffraction limit nor DoF. Go back the the Cambridge Colour page and then go to their section on the DoF calculator for an explanation.

Take a sharp image with a pixel resolution of X. Then take the same image with the same format with a pixel resolution of 2x. Do you think the 2X image will be softer because it is divided into more pixels? The issue is how our eyes perceive an image. A 100% monitor view is not a real world viewing condition and image sharpness in terms of diffraction limit cannot be judged at 100%. My 645D sensor is 33x44mm, do you really think you can judge acceptable sharpness (a technical criteria based on CoC) by looking at a 9mm x 5mm section of the sensor on a 24" monitor, because that is what a 100% view is with a 645D. So as pixel resolution has increased, photographers have been judging images from higher and higher magnifications. Put a microscope on a piece of film and you can see the grain, but you cannot judge image sharpness.

I am sorry, but the theories involving diffraction limit have not changed. Diffraction limit and DoF is not impacted by pixel pitch. It never has. This myth has come from pixel peeping and really not understanding how images are perceived.

BTW, it is not more difficult to focus with a camera with 12MP as it is with 36MP (the focal plane is the focal plane and DoF is not changing). If your lenses create a sharp on a 12MP camera, the images from a 36MP camera will look just as sharp with the same lens--dividing an image into more pixels does not cause an image to be softer. You may be more more aware of the shortcomings at 100% monitor view, but that is not a real viewing condition--look at an image at 200% if you want to see the image softer or 50% if you want to see it sharper.

Here is a paper on depth of field by Zeiss.

http://www.zeiss.com/c12567a8003b8b6f/embedtitelintern/cln_35_bokeh_en/$file/cln35_bokeh_en.pdf