09-25-2011, 09:15 PM
[quote name='Brightcolours' timestamp='1316954803' post='11838']
You are right in that 512nm is still in the green spectrum, I remembered the ranges wrong. It sits already very close to cyan though.
I am unfamiliar with the term air images... Sorry for that, can you explain what it means?
Regardless of what colour 512nm is, monochromatic light says very little about when you start to see diffraction taking its toll resolution wise in normal visible spectrum (white) light, though. With a 5D mk II for instance, you already see less resolution at f8 when you measure with black/white targets in normal light spectrum (see the dpreview tests for instance).
![[Image: mtf.png]](http://www.opticallimits.com/images/8Reviews/lenses/canon_17_4_5d/mtf.png)
![[Image: mtf.png]](http://www.opticallimits.com/images/8Reviews/lenses/canon_70200_28is2/mtf.png)
And that is at a pixel pitch of 6.4µm, twice that of the pixel pitch you stated for the Nikkor 1 10mp sensor?
And I still am puzzled about the very high numbers you mention (400 lp/mm)... As far as I know, the better normal photographic lenses (for DSLRs) reach a bit above 100 lp/mm. Very specialized lenses like my ultra micro nikkor 55mm f2 can reach much higher, but only for monochromatic light (e-line, a green), only in a very limited image circle (12mm with that lens) and only at a specified magnification level (1:4, object distance 218.18mm). And then still it only reaches 500 lp/mm, with monochromatic light, and that 12mm image circle. At f2.
And this very specialized lens is then diffraction limited at f4.
[/quote]
Air images, I use that term because I don't really know the exact English terminology for this. Effectively, by using an optical bench, it is possible to measure lens resolution directly without the interference caused by a medium like film or a sensor. You are essentially looking at virtual images that way. Using the Rayleigh criterion, a world class lens will in that case be diffraction limited at around F/4, or reach 400 lp/mm, the diffraction limit at that aperture. The 100 lp/mm resolution generally is at MTF-50 (diffraction limit of 190 lp/mm) in combination with high resolution film or a high resolution sensor, while the Rayleigh criterion effectively uses MTF-9.
As to the lightwave frequency selected: this is an international standard, not my choice, and just represents the broader spectrum. If you consider that a little more than a wavelength width of light is enough to be able to distinguish a line, it doesn't really matter all that much which specific wavelength you choose. 0.0005 or 0.000512 was chosen as a good average, representative wavelength. I can't help it that that and the Rayleigh criterion were chosen for the definition of diffraction limits in photographic lenses <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='
' />. It was the choice of some standards committee. And yes, it will likely be a little more with shorter, and a little less with longer wavelengths. On average the wavelength chosen always appeared to be a very good measure, however.
As to the above quoted MTF charts: of course you will see less resolution at F/8. As I indicated, the diffraction limit at a particular aperture and the Rayleigh criterion, may be approached by the simplified formula 1600/N lp/mm, where N is the f-stop. IOW, for F/4 the diffraction limit is 1600/4 = 400 lp/mm, and for F/8 it is 1600/8 = 200 lp/mm, or literally only half that of th ediffraction and posisble resolution at F/4.
Regarding the values in the MTF charts, the value of lw/ih needs to be converted to lp first, which is the absolute value divided by 2, and next that value needs to be divided by the sensor height in mm again, to get the resolution in lp/mm. However, as indicated, somehow Imatest seems to coax more resolution from the lens/sensor combination than seems possible theoretically.
When calculating the resolution of the 5D II sensor by itself, one gets a value of 78 lp/mm (i.e, Nyquist frequency). In theory, it is therefore not possible to get a resolution at F/4 of more than 65 lp/mm (fractions don't make sense here, so rounded to lower whole number). However, if you look at the results obtained in the MTFs, one finds that these values are consistently higher than expected - somehow Imatest seems to do some trickery here I don't fathom, and which you will not see from real life results either.
Regarding the Nikkor lens you mention: indeed at F/2 the diffraction limit is 800 lp/mm (1600/2), which means the lens is not diffraction limited at F/2. It likely is diffraction limited at F/4 indeed, maybe at F/3.5. IOW, it may reach higher resolution at larger apertures, but that doesn't mean it is diffraction limited at those larger apertures. Even so, 500 lp/mm is very, very good indeed, as generally world class lenses don't really get any better at apertures larger than F/4, due to the increasing effect of optical aberrations, uncorrigable to a larger degree than at smaller apertures. I'd guess that for a complete broad spectrum of daylight, the Nikkor would likely also have an average resolution of about 400-450 lp/mm when being diffraction limited, and even for F/2 or F/2.8.
Here is a table of diffraction limits, BTW:
[ATTACHMENT NOT FOUND]
HTH, kind regards, Wim
You are right in that 512nm is still in the green spectrum, I remembered the ranges wrong. It sits already very close to cyan though.
I am unfamiliar with the term air images... Sorry for that, can you explain what it means?
Regardless of what colour 512nm is, monochromatic light says very little about when you start to see diffraction taking its toll resolution wise in normal visible spectrum (white) light, though. With a 5D mk II for instance, you already see less resolution at f8 when you measure with black/white targets in normal light spectrum (see the dpreview tests for instance).
And that is at a pixel pitch of 6.4µm, twice that of the pixel pitch you stated for the Nikkor 1 10mp sensor?
And I still am puzzled about the very high numbers you mention (400 lp/mm)... As far as I know, the better normal photographic lenses (for DSLRs) reach a bit above 100 lp/mm. Very specialized lenses like my ultra micro nikkor 55mm f2 can reach much higher, but only for monochromatic light (e-line, a green), only in a very limited image circle (12mm with that lens) and only at a specified magnification level (1:4, object distance 218.18mm). And then still it only reaches 500 lp/mm, with monochromatic light, and that 12mm image circle. At f2.
And this very specialized lens is then diffraction limited at f4.
[/quote]
Air images, I use that term because I don't really know the exact English terminology for this. Effectively, by using an optical bench, it is possible to measure lens resolution directly without the interference caused by a medium like film or a sensor. You are essentially looking at virtual images that way. Using the Rayleigh criterion, a world class lens will in that case be diffraction limited at around F/4, or reach 400 lp/mm, the diffraction limit at that aperture. The 100 lp/mm resolution generally is at MTF-50 (diffraction limit of 190 lp/mm) in combination with high resolution film or a high resolution sensor, while the Rayleigh criterion effectively uses MTF-9.
As to the lightwave frequency selected: this is an international standard, not my choice, and just represents the broader spectrum. If you consider that a little more than a wavelength width of light is enough to be able to distinguish a line, it doesn't really matter all that much which specific wavelength you choose. 0.0005 or 0.000512 was chosen as a good average, representative wavelength. I can't help it that that and the Rayleigh criterion were chosen for the definition of diffraction limits in photographic lenses <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='
' />. It was the choice of some standards committee. And yes, it will likely be a little more with shorter, and a little less with longer wavelengths. On average the wavelength chosen always appeared to be a very good measure, however.As to the above quoted MTF charts: of course you will see less resolution at F/8. As I indicated, the diffraction limit at a particular aperture and the Rayleigh criterion, may be approached by the simplified formula 1600/N lp/mm, where N is the f-stop. IOW, for F/4 the diffraction limit is 1600/4 = 400 lp/mm, and for F/8 it is 1600/8 = 200 lp/mm, or literally only half that of th ediffraction and posisble resolution at F/4.
Regarding the values in the MTF charts, the value of lw/ih needs to be converted to lp first, which is the absolute value divided by 2, and next that value needs to be divided by the sensor height in mm again, to get the resolution in lp/mm. However, as indicated, somehow Imatest seems to coax more resolution from the lens/sensor combination than seems possible theoretically.
When calculating the resolution of the 5D II sensor by itself, one gets a value of 78 lp/mm (i.e, Nyquist frequency). In theory, it is therefore not possible to get a resolution at F/4 of more than 65 lp/mm (fractions don't make sense here, so rounded to lower whole number). However, if you look at the results obtained in the MTFs, one finds that these values are consistently higher than expected - somehow Imatest seems to do some trickery here I don't fathom, and which you will not see from real life results either.
Regarding the Nikkor lens you mention: indeed at F/2 the diffraction limit is 800 lp/mm (1600/2), which means the lens is not diffraction limited at F/2. It likely is diffraction limited at F/4 indeed, maybe at F/3.5. IOW, it may reach higher resolution at larger apertures, but that doesn't mean it is diffraction limited at those larger apertures. Even so, 500 lp/mm is very, very good indeed, as generally world class lenses don't really get any better at apertures larger than F/4, due to the increasing effect of optical aberrations, uncorrigable to a larger degree than at smaller apertures. I'd guess that for a complete broad spectrum of daylight, the Nikkor would likely also have an average resolution of about 400-450 lp/mm when being diffraction limited, and even for F/2 or F/2.8.
Here is a table of diffraction limits, BTW:
[ATTACHMENT NOT FOUND]
HTH, kind regards, Wim
Gear: Canon EOS R with 3 primes and 2 zooms, 4 EF-R adapters, Canon EOS 5 (analog), 9 Canon EF primes, a lone Canon EF zoom, 2 extenders, 2 converters, tubes; Olympus OM-D 1 Mk II & Pen F with 12 primes, 6 zooms, and 3 Metabones EF-MFT adapters ....