>>3617762Excuse me, but that doesn't contradict anything I'd stated?
0.9x0.5x0.8x0.65=0.234, just as the example said. Thus this is in line with my:
>MTF of the system is reduced the more elements you add.>The total MTF of the system is lower than MTF of either of the elements.But sure, nothing ever reaches that max value, camera and lens are essentially are the receiver and antenna for EM waves in the optical spectrum, and dealing with imperfect transmissions is the crux of signal processing. But trying to sell people that going from 0.9 to, say, 0.95 will be noticeable is just marketing. Heck add an optically flawless lens and the result will be 0.26, compared to 0.23, was it worth it?
>Play with those numbers a bitGreat idea, let's see some real world examples. A quick lookup yielded pic related (from
https://www.osapublishing.org/DirectPDFAccess/A5B96BE6-9531-528A-75D61F6919846973_174902/oe-16-24-20047.pdf?da=1&id=174902&seq=0&mobile=no ), these are the values they got for a sensor with a pixel pitch of 6 microns. Similar to that of a 24 megapixel ff sensor. It doesn't drop below 0.9 until at least 20 lines per mm. In the same paper, they also present a 2.2 micron monochrome sensor that gets maybe 0.96 at 20 lines/mm. Let's say that Canon's 50 megapixel sensor is somewhere between the two, so 0.93 at 20lines/mm. Grab some expensive lens, say, Canon's 85mm f/1.2 L II, and see that at 20 lines/mm it doesn't get above 0.7 in the center (and 0.4 in corners), source:
the-digital-picture.com.
So, our system with a 24Mp sensor would get 0.63 (0.36 corners), and a 50Mp sensor gets 0.65 (0.37 corners). So, we doubled the megapixels to get a 1-2% increase in iq. It's not as bad at higher densities though,say, 40l/mm gives 0.6 to a 24, 0.8 to a 50, 0.45 to the lens, giving us MTF(24Mp, 40l/mm)=0.27 and MTF(50Mp, 40l/mm)=0.36 in in the center, but corners get demolished (think sub 0.1), rendering photos useless. Okay that was sort of fun and productive.
