Originally posted by WorksAsIntended This is not true. The diffraction limit is way more complex than it gets cooked down often these days in camera forums.
Almost everything is more complicated than what gets posted on camera forums.
However diffraction limits, just for a general idea can be easily identified from lens charts.
A typical lens chart...
Looking at the DA 70 on a 10 MP camera
It gets better and better until you get to ƒ5.6 then diffraction kicks in.
At 16 MP, same density as a K-1 the diffraction limit would appear to be somewhere between ƒ4 and ƒ5.6. The smaller pixel sites are more affected by diffraction. ƒ8 and ƒ11 are over the diffraction limit and are losing resolution even though without diffraction they'd be the sharpest.
Looking at these two charts you can draw a few conclusions not only about where the diffraction limit is, but see how it is affected by pixel size and location on the sensor, as the edges seem more affected by diffraction than the centres on the smaller pixel cameras, which is no doubt what led to the DFA's made for full frames being so much more heavily corrected.
It's unfortunate Klaus has never done a 24 MP comparison. And if you look through the lens charts it's interesting to see, how different lenses have different characteristics at different MP counts.
So when you're adding in all the things you need to add in, lens design and characteristics etc. etc. be sure and deal with pixel density as well.
These lenses are tested at a specific distance from the target image. What they'd look like tested at different distances from the target is a whole different issue.
Really, for so many reasons, this is not a subject you want to try and analyze mathematically. "I like this lens on this camera." Is all you need. Start trying to explain why and you're in for a world of over simplified information, and possibly misleading information.
Couple that with the fact that a bit of diffraction may make your image more pleasing to look at, and you quickly realize, not knowing what you don't know can quickly make your conclusions meaningless. You have know what the lens does on the camera you use. That's all that's important.
My guess is that the empirical knowledge of lens design contains elements so far removed from what you can establish without years of producing and evaluating lenses, it would make any attempts at mathematical analysis pointless.
Somewhere there was a description of one Pentax lens where the design process was based not on the test charts, but on which variant of the lens produced prints most enjoyed by members of the general public leading to the philosophy "lenses for the way people take pictures, not for the test charts". IN other words the pure test chart lens was not the lens selected for production, based on not as many people liking it's images. I'm fairly certain of a couple things with regards to that. The DFA* 50 1.4 was designed in part with that criteria. The change being they designed a lens to excel both on the test charts and in actual usage. I have no doubt Jun Hirakawa took some of that with him to Tamron, and that you're seeing something similar in the adopted tamron lenses. They have math they aren't explaining to us... and if they did, they'd have to kill us.
Or to return to 1966 philosophy, find the lens you love, buy the camera it goes on.