02-27-2012, 09:34 PM
Hi Frank,
[quote name='Frank' timestamp='1330345797' post='16226']
Thank you very much, Wim. Your explanation is very clear and very comprehensive. I appreiciate it very much. I think I understand it now.
The optical resolution derived by me simply exceeds the Rayleigh limit and this is impossible. [/quote]
Yes, indeed.
The type of analysis used here bears resemblance to those used in image analysis of satellite imaging. It is debatable how useful that is in normal photographic imaging. Personally, I think it is a pity the analysis implemented wasn't a standard MTF-50, or even better, according to the Rayleigh criterion, but hey, at least we get quite consistent results here, compared to other test sites.
It looks like the AA-filter scales down resolution by approximateley 1 f-stop, or around 15% or thereabouts. Note that expected system resolution according to Rayleigh criterion is about 60 lp/mm in this case, which is very good indeed.
To convert the PZ MTF figures to proper system resolution would require implementing a new analysis method and methodology, possibly different software, and reprocessing all test images <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='
' />. I doubt this will happen any time soon, if ever <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />.
I have been working on something to see if I can somehow scale the results, but when compared to other review sites, one finds not only the figures are derived in a completely different manner (and all review sites are different), measurements are also taken at different points in the image, they are not the same from review site to review site, or even system to system on the same site.
One thing is very typical for Photozone: quite often the border and extreme border figures are quite close together compared to other test sites. This seems to be caused by Photozone using measuring points lying furthest out to the corners proper of the sensor. IOW, this provides a very good idea of the possibly worst performance of a lens, even if it generally is a lot better, in real life anyway.
Kind regards, Wim
[quote name='Frank' timestamp='1330345797' post='16226']
Thank you very much, Wim. Your explanation is very clear and very comprehensive. I appreiciate it very much. I think I understand it now.
The optical resolution derived by me simply exceeds the Rayleigh limit and this is impossible. [/quote]
Yes, indeed.
Quote:Now I know that the PZ resolution numbers are not the system resolution, they are just some numbers generated by the software.
The type of analysis used here bears resemblance to those used in image analysis of satellite imaging. It is debatable how useful that is in normal photographic imaging. Personally, I think it is a pity the analysis implemented wasn't a standard MTF-50, or even better, according to the Rayleigh criterion, but hey, at least we get quite consistent results here, compared to other test sites.
Quote: Following your suggestion, I compared the resolution numbers of the Nikon AF-S 35mm f1.4 tested on D3x from PZ and lenstip. At F4, the PZ center resolution is "translated" to 82 lp/mm, while the lenstip resolution is only about 46 lp/mm (MTF-50).
It looks like the AA-filter scales down resolution by approximateley 1 f-stop, or around 15% or thereabouts. Note that expected system resolution according to Rayleigh criterion is about 60 lp/mm in this case, which is very good indeed.
Quote:So, indeed the PZ resolution number is not the system resolution.
I guess there is no simple way to convert the PZ resolution number to the system resolution since otherwise the system resolution will be used in displaying the test result.
Best regards,
Frank
To convert the PZ MTF figures to proper system resolution would require implementing a new analysis method and methodology, possibly different software, and reprocessing all test images <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='
' />. I doubt this will happen any time soon, if ever <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />.I have been working on something to see if I can somehow scale the results, but when compared to other review sites, one finds not only the figures are derived in a completely different manner (and all review sites are different), measurements are also taken at different points in the image, they are not the same from review site to review site, or even system to system on the same site.
One thing is very typical for Photozone: quite often the border and extreme border figures are quite close together compared to other test sites. This seems to be caused by Photozone using measuring points lying furthest out to the corners proper of the sensor. IOW, this provides a very good idea of the possibly worst performance of a lens, even if it generally is a lot better, in real life anyway.
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 ....
