05-10-2011, 09:55 PM
[quote name='arv' timestamp='1305029669' post='8201']
Hi,
IMO all these comparisons are quite useless. First of all, there are measurements errors, both for lenses and cameras. There are limits of tolerance for both. In FAQ Klaus writes about 5% measure's error, so in fact we should talk about confidence intervals, but not about exact numbers. Second (assuming sample variations), usually for 95% confidence interval you need about/more 30 samples (lenses)(well that depends on dispersion, but let it be) and assuming camera's variation - 30 cameras <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/smile.gif' class='bbc_emoticon' alt='
' /> (lets say to evaluate sensor's decentering). This kind of testing might take a while and we can get one review in a couple of years <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/smile.gif' class='bbc_emoticon' alt='' />. The numbers you see represent lens performance, but only represent. Good lens is good, mediocre is mediocre, and that's all.
A.
[/quote]
Very helpful perspective. Thanks
Hi,
IMO all these comparisons are quite useless. First of all, there are measurements errors, both for lenses and cameras. There are limits of tolerance for both. In FAQ Klaus writes about 5% measure's error, so in fact we should talk about confidence intervals, but not about exact numbers. Second (assuming sample variations), usually for 95% confidence interval you need about/more 30 samples (lenses)(well that depends on dispersion, but let it be) and assuming camera's variation - 30 cameras <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/smile.gif' class='bbc_emoticon' alt='
' /> (lets say to evaluate sensor's decentering). This kind of testing might take a while and we can get one review in a couple of years <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/smile.gif' class='bbc_emoticon' alt='' />. The numbers you see represent lens performance, but only represent. Good lens is good, mediocre is mediocre, and that's all.A.
[/quote]
Very helpful perspective. Thanks