09-03-2015, 09:55 PM
The issue of focusing distance comes much more into play with wide angles than telephotos. A tele is going to behave quite similarly at 20ft as it will at 200ft or it will at infinity. A retrofocus wide-angle will usually not behave at 2ft like it does at infinity. So if lenstip is using a 4ft diameter chart, that's a problem. Unless all you ever do it shoot around that focusing distance with your 24mm lenses =) Even 8-10ft would be perfectly sufficient, but target size becomes impractical.
The issue of a coverglass has to do with different cameras. They have a "custom 200MP sensor" which to me is probably a line scanning device. No coverglass is alright, the difference between 0 and 2mm (~canon or so) is not too enormous. If it is more like 4-5mm, that is a big problem as it will inject simply enormous spherical aberration into the equation.
The bigger issue with DxO is the amount of mathematics poorly applied to the data they gather. An MTF bench does a huge amount of math to produce results, but it is all to resolve numbers out of the point spread function. DxO's computations are to calculate the "max" resolution of the camera, which can't be known without a lens attached, weight the lens+camera res against camera+(what lens?), and other simply silly math.
The testing done at OLAF on the ImageMaster is not without its caveats and issues so to speak, but no testing is which is why clearly defined and stated metrology parameters are needed. Sadly, the latter is missing with most sources.
The issue of a coverglass has to do with different cameras. They have a "custom 200MP sensor" which to me is probably a line scanning device. No coverglass is alright, the difference between 0 and 2mm (~canon or so) is not too enormous. If it is more like 4-5mm, that is a big problem as it will inject simply enormous spherical aberration into the equation.
The bigger issue with DxO is the amount of mathematics poorly applied to the data they gather. An MTF bench does a huge amount of math to produce results, but it is all to resolve numbers out of the point spread function. DxO's computations are to calculate the "max" resolution of the camera, which can't be known without a lens attached, weight the lens+camera res against camera+(what lens?), and other simply silly math.
The testing done at OLAF on the ImageMaster is not without its caveats and issues so to speak, but no testing is which is why clearly defined and stated metrology parameters are needed. Sadly, the latter is missing with most sources.