[Guest]

Seems to me that main issues with at least M-lenses sharpness lies in AA-filter+IR filter thickness.

Both Ricoh M module & Leica do not have AA-filter & at least Leica cameras struggle to have IR filter with minimal thickness possible.



It is possible, that even for mirrorless cameras this extra glass could be at least partially compensated in lens design.



Read this info by lens designer of Coastal Optics 60mm f/4 UV-VIS-IR APO Macro:

http://diglloyd.com/articles/CoastalOptics60f4/designer-notes.html

The second issue relates to aberration correction. Any time you place a plane-parallel piece of glass in the path of a converging light beam you introduce aberrations. Virtually all digital cameras have a cover glass of some sort. Even if you remove the filter pack there is still a thin coverslip that protects the actual pixel structures. The vast majority of digital cameras also include a filter in addition to this coverslip. In order to achieve the best possible performance in a lens to be used with digital cameras it is therefore necessary to include the effects of all this glass near the image plane, and to counteract it by proper design of the lens.



When designing the 60mm f/4 I called several people who were doing IR and UVIR camera conversions, and was told that DSLR filter thicknesses ranged from ~1mm to more than 3mm. There is no standard thickness, unfortunately. However, I had to pick something, and 2mm seemed like a reasonable compromise. It seems that this choice might be a very good match for a standard Canon 5D, since the filter pack is 1.45mm thick and the coverslip is probably ~0.5mm thick. Making a small error in filter thickness has a fairly minor effect on the aberrations, and even omitting it altogether will not cause a drastic reduction in performance. After all, any lens designed for film cameras will automatically have aberrations induced by the sensor glass when used on a digital camera, but very few people notice this effect.



The sensor glass aberration effect is much greater when the exit pupil is shorter. Quasi-symmetrical large format lenses such as the Rodenstock HR series, which have a very short exit pupil distance relative to the image diagonal, really must account for the sensor glass in order to maintain good performance in the corners. In our case the exit pupil is much further away, and my decision to account for the sensor glass is based on a simple desire to achieve the finest possible image quality. From the beginning my goal has been to achieve not just ultra-broadband correction, but also to achieve better visible band performance than any other manufacturer.