If we assume a perfect lens, at f/2.8 and 50% contrast, you can resolve about 247 lp/mm...
How do you determine that number? The common formula of "1600/f-stop = lp/mm" means a perfect lens at f/2.8 would resolve about 571 lp/mm, over twice your figure of 247 lp/mm.
I am basing my numbers on an MTF of 50% (hence the "at f/2.8 and 50% contrast"). I believe the 1600/f-stop is based on rayleigh, or an MTF of 9%. MTF 50% is commonly used with photography, where as Rayleigh/MTF 9% is usually used in reference to human visual acuity (and then, usually referring to our ability to barely resolve two closely spaced points of dim light on a black background...i.e. stars). I prefer MTF 50%, as that relates better to the clear, sharply defined kind of detail people prefer in their photography, where as MTF 9% might apply assuming you were ok with detail of extremely low contrast (i.e. barely discernible differences.) To be exact, MTF at 9% is about 532 lp/mm, so the 1600/f-stop is really pushing it for cameras.
Once you get to that level, or when you start talking about MTF 0% (which technically is just barely above 0%), there are only a few cases where such low contrast is applicable. The most notable being the detection of binary stars from a what otherwise resolves as a single point-light source in astronomy & astrophotography...at just above MTF 0% you can detect whether a resolved point in a photograph represents a single star or a binary (or even ternary) star based on how the diffraction pattern presents.
Currently, there isn't a commercial-grade camera that can even come close to resolving that much detail, it usually requires high resolution, scientific-grade cooled CCD's and appropriate detection software. You can also use superresolution techniques with commercial-grade gear to produce images that can then be processed to identify binary stars at MTF 0%, which is something the amateur binary star discoverer can do these days if they wish, although its not as effective (superresolution is still a newer technique, and its not guaranteed to leave the presentation of a multi-star airy disc in tact in all cases.) The best consumer-grade sensors on the planet are barely capable of 130lp/mm, and when you factor in the nature of multi-component optical systems, the actual final spatial resolution of a whole camera system tends to be quite a bit lower than that of the highest resolution component (be it lens, sensor, whatever.) So...MTF 50%...in my opinion, it is more realistic for real-world photographers.
As a side note, modern lenses are not restricted to projecting light at any given contrast level. Measurements of resolution, in the form of MTF, determine spatial resolution at a given contrast level...the contrast level is a factor of the measurement, but it is not a limitation of the lens itself. As such, there is nothing to prevent a LENS from projecting an image at any contrast level...0% to 100%. It is entirely possible to project an extremely fine white line on a black background that is 1/(571*2)mm thick (0.000876mm) with a lens. The contrast of that line could be extremely low (so blurred that it barely registers more than pure black)...even below the level at which the human eye can detect (which would be 9% contrast), and well below the level at which the best modern DSLR or even medium format sensors could resolve in any meaningful sense.
There are some films that are capable of resolving far more than the best sensors today, such as a Zeiss film capable of resolving about 400lp/mm at ISO 25 (which, as far as I know, was only ever used for the purposes of testing a (possibly..it may still exist) short-lived and somewhat legendary...in certain circles...400lp/mm ultra fast lens.) It should be noted, however, that the nature of film and a very meticulous and expensive manufacturing process makes it a bit easier to support such incredible spatial resolutions (which would only be possible at very wide apertures, or at barely discernible contrast levels and narrower apertures). Most digital sensors follow a bayer array design, and usually have low-pass filters in front of the sensor. That puts a hard limit on the amount of resolution you can achieve digitally. Canon has claimed their newer L-series lenses (not exactly sure what "newer" means...from the time I read that, it was around 2008 or 2009, so perhaps within a few years of that date) are capable of resolving 45mp worth of full-frame CMOS sensor resolution...which pipes in at arounb 113-116lp/mm (f/5.6 or wider, as any narrower than that and diffraction limits your resolution.) A sensor might barely be able to resolve a
Logged