Originally posted by Aristophanes Aristophanes, thank you very much, seriously interesting blog.
His results quantitatively show that the use of good adapters is unlikely to be problematic for Q system.
The height of 1/1.7" sensor in landscape is (-4.7, +4.7) degrees and the full image circle (-7.8, 7.8) on his plots. For 1/2.3" sensor they're (-3.7, +3.7) and (-6.3, +6.3).
The data is only measured at 0, +-4, +-8 and +-12 degrees but I don't see any indication that the image quality becomes unacceptable inside (-4, +4) degrees, and even at +-8 degrees some example don't show much degradation.
And this is a 35mm lens. In general the larger the focal length, the less you're affected by the relative tilt between the image and the sensor plane. (That's not to say that it doesn't matter for longer lenses at all, though, it all depends on what you want to get.)
I have a great respect for people like this guy who perform meaningful tests AND publish the results. Just like Sandy Hancock who took moon pictures with DA* 300, one with Q and the other with K-5 IIs and put them on flickr, or just like reh321 who did Q7 and K-30 comparison in
this forum post (though it seems like his pictures are gone). And of course the great thing is that these results are consistent with each other.
Because of these people we can ask meaningful questions and answer them, without making sweeping statements like "adapters don't work", "Of course they work. but they all degrade IQ.":
- Is the degradation due to supposedly non-parallel flange surfaces or other imperfection of adapters measurable within Q image circle using some equipment?
- Is the degradation due to adapter perceptible when you take a picture using Q?
- Looking at his data as well as tests done by people like Sandy and reh321, and many pictures made using adapters including many of mine, I'd say "probably not" for majority of cases.
- But it's possible (not impossible) that it's perceptible for certain type of pictures, e.g. stars and test charts and whatnot where everything in the picture is supposed to be on the same image plane, corner to corner. I'd like to see such an example.
- Does it matter as far as pictures taken by Q and adapters are concerned?
- No if not perceptible.
- Depends if perceptible.
- Yes if you value perfect corner to corner sharpness, e.g. star pictures. Does anybody have an example?
- No otherwise, maybe?