On a practical level, Herb is correct. Technically, the diffraction limits of an optical system are determined by the pixel size in the sensor and the wavelength of the light. Here is a table I put together for this purpose based on the laws of diffraction. The smaller pixel = D300 and the larger pixel is either the D70s or the D3 ... I don't remember witch I was using. The most reliable result is the average of the 1-pixel number.
molson wrote:
That's correct, diffraction is a characteristic of the lens, not the sensor.
Diffraction is a function of aperture not focal length. If there is lens dependence, then that means there is something else in the optics causing an issue (i.e. perfect optics are diffraction limited). And, diffraction issues most definitely depends on the sensor. Optics 101.
OK, I'm confused. Diffraction is the bending of light. So how does a sensor bend light? I thought it is a product of the lens. The sensor size only impacts the lens diffraction performance.
Andre Labonte wrote:
Diffraction is a function of aperture not focal length. If there is lens dependence, then that means there is something else in the optics causing an issue (i.e. perfect optics are diffraction limited). And, diffraction issues most definitely depends on the sensor. Optics 101.
The bigger the sensor the later diffraction has effect. The same lens will be diffraction limited at a smaller aperture in a bigger sensor. I think the accepted result is that FX has a one stop advantage over DX for diffraction. So if your image becomes smooth mash at f/11 on your D300, it will still be chunky mashed potatoes on the D3 and will be back to smooth mash at f/16.
Furthermore, different lenses have slightly different diffraction limits, depending on their design, after all, they're not all equally sharp to begin with.
Jul 17, 2008 at 09:12 AM
Andre Labonte Offline • • • • • • Upload & Sell: Off
panos.v wrote:
The bigger the sensor the later diffraction has effect. The same lens will be diffraction limited at a smaller aperture in a bigger sensor. I think the accepted result is that FX has a one stop advantage over DX for diffraction. So if your image becomes smooth mash at f/11 on your D300, it will still be chunky mashed potatoes on the D3 and will be back to smooth mash at f/16.
Furthermore, different lenses have slightly different diffraction limits, depending on their design, after all, they're not all equally sharp to begin with.
That is correct. For a given Mega-pixel value, sensor size matters because of its effect on pixel size. Ultimately, it is the size of the pixels. You will start to see the effects of diffraction sooner with smaller pixels. So while smaller pixels for a given format (i.e. more MP) provide greater resolution, that greater resolution can only be realized if the chosen aperture (actually it's f-stop) can support that resolution. This of course assumes a print size that allows you to see the effects as well.
As for the effects of lenses, again, if the lenses are not diffraction limited, the lens will matter. Most lenses are not diffraction limited, though some of the new lenses are approaching that limit.
Andre Labonte wrote:
That is correct. For a given Mega-pixel value, sensor size matters because of its effect on pixel size. Ultimately, it is the size of the pixels...
Not really.
Well, it is, but only if you're peaking at 100% on the monitor. Or, print really big. For reasonably sized prints the determining factor is the CoC, which is larger than the pixel size (unless you print really big). Thus, you can close the aperture more than calculated based on pixel size.
Jul 17, 2008 at 03:51 PM
Andre Labonte Offline • • • • • • Upload & Sell: Off
Well, it is, but only if you're peaking at 100% on the monitor. Or, print really big. For reasonably sized prints the determining factor is the CoC, which is larger than the pixel size (unless you print really big). Thus, you can close the aperture more than calculated based on pixel size.
I did say that print size had to be large enough to see the effects AND that I was speaking of diffraction limited optics. I don't think it takes pixel peeping at 100% to see it either. But you do bring up a good point, so many people worry about going past the diffraction limit without realizing that their assumed CoC is actaully larger than the AD. And yet, the image looks good.
I did say that print size had to be large enough to see the effects AND that I was speaking of diffraction limited optics. I don't think it takes pixel peeping at 100% to see it either. But you do bring up a good point, so many people worry about going past the diffraction limit without realizing that their assumed CoC is actaully larger than the AD. And yet, the image looks good.