06-15-2011, 02:31 AM
[quote name='guy_incognito' timestamp='1308082209' post='9210']
photozone has been posting MTF50 data in the center and borders. It would be nice to see MTF25, MTF75 or other values from several spots across the frame:
MTF curves are never linear. If you just look at MTF curves released by manufacturers, sometimes the curve drops off dramatically at the border, while other times, it drops off gradually. In the first case, image quality only on the borders would be compromised. In the latter case, image quality woud suffer across the frame. Also, the choice of "50%" from the explanation on imatest is a bit dubious and ad hoc. The author mentions something about printers... MTF 25 or something close to it could tell us about "resolution" as in the ability to *resolve* detail, while MTF 75 or something close could tell us about "overall contrast." Note that the graph with MTF on the y-axis and LW/PH on the x-axis is also never quite linear. Thus, the MTF50 numbers don't tell us the whole story.
These are just some words to encourage including *more* charts and data in future reviews...and maybe a fancy Java applet to navigate through this data. <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/smile.gif' class='bbc_emoticon' alt='
' />
[/quote]
+1
This wasn't really an issue in the earlier days when testing was mostly full-frame lenses on APS-C sensors. Nor when even recently when testing 'u'ber-lenses (i.e. Canon L and Zeiss ZE/ZF) on full-frame sensors. They are relatively huge chunks of glass relative to sensor size.
But testing either "typical" compact-prime wide-angles or large range zoom lenses on either full frame (i.e. Nikkor 28/2.8, 24/2.8, Ultron 20/4) or APS-C (i.e.Pentax DA15/4, DA18-135) there is an irrational expectation of performance in the corners.
On lenses with weak corners, I would like to see additional data points included at the golden-triangle intersection.
photozone has been posting MTF50 data in the center and borders. It would be nice to see MTF25, MTF75 or other values from several spots across the frame:
MTF curves are never linear. If you just look at MTF curves released by manufacturers, sometimes the curve drops off dramatically at the border, while other times, it drops off gradually. In the first case, image quality only on the borders would be compromised. In the latter case, image quality woud suffer across the frame. Also, the choice of "50%" from the explanation on imatest is a bit dubious and ad hoc. The author mentions something about printers... MTF 25 or something close to it could tell us about "resolution" as in the ability to *resolve* detail, while MTF 75 or something close could tell us about "overall contrast." Note that the graph with MTF on the y-axis and LW/PH on the x-axis is also never quite linear. Thus, the MTF50 numbers don't tell us the whole story.
These are just some words to encourage including *more* charts and data in future reviews...and maybe a fancy Java applet to navigate through this data. <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/smile.gif' class='bbc_emoticon' alt='
' />[/quote]
+1
This wasn't really an issue in the earlier days when testing was mostly full-frame lenses on APS-C sensors. Nor when even recently when testing 'u'ber-lenses (i.e. Canon L and Zeiss ZE/ZF) on full-frame sensors. They are relatively huge chunks of glass relative to sensor size.
But testing either "typical" compact-prime wide-angles or large range zoom lenses on either full frame (i.e. Nikkor 28/2.8, 24/2.8, Ultron 20/4) or APS-C (i.e.Pentax DA15/4, DA18-135) there is an irrational expectation of performance in the corners.
On lenses with weak corners, I would like to see additional data points included at the golden-triangle intersection.
/Dave
http://dave9t5.zenfolio.com
http://dave9t5.zenfolio.com