Since my last post on image resolution systems and Modulation Transfer Function (MTF) functions I opted to create larger Koren targets that would be useful for better evaluating long lenses and telescope optical tubes. Here are the targets, one four times and one two times the original size as well as a larger ISO 12233 chart increased by 50% or 300mm page height.
I intended to place this target at the recommended calculated distance from the Koren web site of approximately 400 ft. (122m) but the site I chose ran out of room so was limited to a distance of about 350 ft. (107m) but this was not a problem since I calculated the angles rather than use the industry standard lp/mm for than actual minimum resolving angles.
My previous calculation for minimum resolving angle for the 9.25 EdgeHD optical tube coupled with the Nikon D3s were very close, I calculated the minimum resolving angle of about 4.5 arc-seconds. Any details less than this angle are aliasing and moire effects and corresponds to an MTF of about 10%, the 50% MTF translates to about 15 arc-seconds.
I also tested the same setup with the D300 and discovered some interesting results. The image from the D3s actually looked slightly sharper while the D300 image looked to be somewhat out of focus. I shot about twenty separate images refocusing between each shot. I used the mirror lockup. I then picked the sharpest image to analyze. Here is the image from the D300 for comparison:
I then manually measured and logged the data in a spreadsheet. I also incorporated the normalization recommended on the Koren web site using the extrema values for the black and white pixel values to compensate for the gamma. This assures the graph uses the full values from 0% to 100%. Interesting that the D300 MTF beat the D3s MTF up to the 50% MTF point. The 50% MTF point was actually the crossover point where the D3s results were actually superior. This clearly demonstrates the importance of the areas between 50% and 10% MTF in how the images are evaluated qualitatively or aesthetically. Here is the graph: (corrected 2013-03-25)
(Added 2013-03-25) This perceived superiority of the sharpness, in this particular case, is due to the fact that the above images consist of mainly fine line details. Traditionally small pixel size sensors will generally be perceived to be sharper because most of the information, in the vast majority of images, consists of lower spacial frequency data.
In those cases the area under the 100% to 50% graph will dominate the aesthetics or perceived qualities of the images. This was one of the many contributions to the field of imaging resolution analysis by Erich Heynacher from Zeiss.
The larger pixel sensors generally tend to be more susceptible to distracting moire and aliasing noise since they have lower spacial frequency gratings that tend to interact or modulate the signal because these frequencies tend to be in the range of sharp focus. The purpose of anti-aliasing filters is to defocus these frequencies.
In conclusion these results fully compliment my past evaluations of the Celestron 9.25 EdgeHD tube where I have concluded that this particular optical tube is substandard from the perspective of high quality astrophotography. Since My Nikon 70-200mm 70mm diameter telephoto lens is very close to the resolving power and its f/2.8 aperture actually allows for shorter exposures than the f/10 Celestron tube, even with its potentially greater light gathering diameter.
(added 2013-03-29) The measured absolute minimum resolving power of the Nikon 70-200mm f/2.8 lens with the 1.4x teleconverter and D3s was actually about 12.8 arc-minutes, which is pixel size limited, it is NOT actually diffraction limited by the lens optics.
If you are seriously considering an astrophotography lens look elsewhere and seriously consider an APO refractor instead. For visual use this Celestron tube is satisfactory but a standard non EdgeHD SCT scope would probably perform much better and also be cheaper in the long run.