02-26-2015, 11:10 PM
Quote:A sensor with 55% more pixels inflates the numbers for sony's sensor.
The format for imatest results is lw/ph, if the lens and detector were perfect you would see a 1:1 correlation between lw/ph and the longest edge of the sensor in px.
The 5D2 has 5600px long edge. Peak imatest numbers for it are about 3850lw/ph but this is an exceptionally good lens performance (24L II at f/4) and the "true max" is about 3700. Calculate the sensor's "resolution factor" here = 66% which is very expected.
A7r -> 7600px long edge. Peak resolution is about 4700lw/ph or 61% which is a bit below expectation.
We may bump canon's numbers by the % difference between the two in linear resolution = 7600/5600 = 35%.
We may also add 10-15% on top of that for the removal of the AA filter but some may disagree with this so I will do it separately.
16mm comparison:
Center:
16-35/4L : 4800lw/ph
16-35/4S : 4500lw/ph
Corner:
16-35/4L : 3500lw/ph
16-35/4S : 3199lw/ph
AA-filter compensated:
Center:
16-35/4L : 5280lw/ph
16-35/4S : 3850lw/ph
Corner:
16-35/4L : 3500lw/ph
16-35/4S : 3199lw/ph
Comparisons are pretty similar at other focal lengths.
Canon also achieves this resolution with 50% less vignetting. Not to mention the (astronomically) worse constraints because efl << bfl for the canon design.
CAs on the canon are also about 1/2px or about (36/5600) = 1/2 * 6.4um = about 3.2um across the range.
Sony has about (36/7600) = 4.7um pixels * 1-2px of lateral CA = at least 4-8um of CA.
I'm no expert in this but could you really bump resolution figures without knowing if the lens is the limiting factor? And what does Photozone's "excellent line"mean (which the Canon lens never passes outside the center).
It sounds like you're really want the Canon to be better...