09-13-2011, 11:10 AM
[quote name='jenbenn' timestamp='1315910508' post='11541']
Ever tried to duplicate their measurments in real life? Dxo is the most sientific yet most unreliable optical testing bench I have ever seen. Example? Check this test of the Canon 16-35mm lens out:
[url="http://www.dxomark.com/index.php/Lenses/Camera-Lens-Database/Canon/EF16-35mm-F2.8L-II-USM/%28camera%29/436#div1anchor"]http://www.dxomark.c.../436#div1anchor[/url] ( Got to "field map" and choose focal length 16mm and aperture accordingly)
According to their results, the borders at 16mm are sharpest at f/2.8. Stopping down the lens leads to increasingly soft borders. At 5f/5.6 the lens is claimed to have half the border resolution compared to 2.8. What a joke. Anybody who has ever shot with the lens knows that it becomes sharper on stopping down. (Although the corners certainly never become totally sharp).
Same with their camera noise measurments. Because they dont put the noise in relation to pixel resolution, it is completly impossible to compare one camera to another. E.g. If you print an image from 24MP camera with medium noise at 13 x18" it may look the same or better than an image printed at the same size from a camera with 15Mp but lower noise.
[/quote]
Interesting... and only for 16mm FL we can see that odd result... For example @20mm wide open, corners are worse than f/5.6 - 8 interval (as expected).
This can be clearly seen by checking the "global map" / corners... Do they have an explanation for that?
Ever tried to duplicate their measurments in real life? Dxo is the most sientific yet most unreliable optical testing bench I have ever seen. Example? Check this test of the Canon 16-35mm lens out:
[url="http://www.dxomark.com/index.php/Lenses/Camera-Lens-Database/Canon/EF16-35mm-F2.8L-II-USM/%28camera%29/436#div1anchor"]http://www.dxomark.c.../436#div1anchor[/url] ( Got to "field map" and choose focal length 16mm and aperture accordingly)
According to their results, the borders at 16mm are sharpest at f/2.8. Stopping down the lens leads to increasingly soft borders. At 5f/5.6 the lens is claimed to have half the border resolution compared to 2.8. What a joke. Anybody who has ever shot with the lens knows that it becomes sharper on stopping down. (Although the corners certainly never become totally sharp).
Same with their camera noise measurments. Because they dont put the noise in relation to pixel resolution, it is completly impossible to compare one camera to another. E.g. If you print an image from 24MP camera with medium noise at 13 x18" it may look the same or better than an image printed at the same size from a camera with 15Mp but lower noise.
[/quote]
Interesting... and only for 16mm FL we can see that odd result... For example @20mm wide open, corners are worse than f/5.6 - 8 interval (as expected).
This can be clearly seen by checking the "global map" / corners... Do they have an explanation for that?