Originally posted by normhead The fact that there can be a 20% margin of error just means no results can be taken as definitive.
This is pretty much my point, every test published from a single sample is just one data point. Not definitive, unless
multiple lenses have been tested on exactly the same testing rig.
Also many lens testers to not mention the distance from the resolution target the sample was taken. Many long zoom lenses do not resolve well at closer focus distances case in point:
100% crops - One of these images was taken from the excellent Sigma 100-300mm f/4 Zoom - compared to a dedicated macro 180mm prime. The differences are visible, but in a 11"X14" print who would notice?
Also, Long lenses are a
PITA to test on optical benches due to the distances needed to achieve true infinity focus*, and if the lens extends at all at that setting** then there are stability and alignment issues that need to be taken into account.
*Most benches only extend about 300mm, fortunately mine goes out to 1200mm because it was actually designed for testing LF lenses. The problem is it needs an absolutely immovable platform to deliver accurate and repeatable results, there can be no vibrations.
** depending on the design of the bench Zoom creep/focus shifts can be rather problematic.
---------- Post added 2020-03-31 at 12:44 AM ----------
Originally posted by Kunzite Of course it can. But every single time?
Show me the data.
Originally posted by BigMackCam I get that sensor-stack thickness could impact test results on wider angle lenses, but on a tele zoom I'd have thought it would have negligible impact.
It definitely does affect wide angle lenses, telephoto lenses can be affected in more subtle ways but the sensor stack the telecentricity of the lenses optical design matters considerably.