« on: October 23, 2012, 05:12:59 AM »
As for the question posed by PerfectSage, there does appear to be a real and practical answer, or at least rule of thumb, which would guide one towards the goal of advantaging all of that 116 lp/mm resolving power of the 7D sensor, and that is to choose optics that will present an image to the sensor with enough inherent detail. if the source image truly does not contain the detail, the sensor will not find any that isn't there. Whether or not that goal is a good one or not can be debated of course
Technically speaking, there is an asymptotic relationship in terms of spatial resolution. You can never actually achieve the same spatial resolution as the highest resolving component in an optical system. As you approach it, you begin to experience diminishing returns. Lets say you have a lens capable of resolving 86lp/mm. Nothing you ever do can ever allow you to resolve 86.1lp/mm...your upper bound is the resolution of the lens itself. At best, you could reach 85.99999999999... lp/mm, assuming you had a sensor with literally infinite resolution. You would need something like an f/0.3 lens to resolve around 115lp/mm of resolution, and approach the 116lp/mm of the 7D. Total "system spatial resolution" is derived from the RMS of the "blur circle" of each component in an optical system. The size of the airy disc at a given aperture in the lens, blur introduced by any and all TC's, the size of a pixel in the sensor, and if you want to get really accurate, the size of the blur introduced by low-pass and IR cut filters. Taking the RMS of each of those will give the the size of the blurry disc of a single point light source resolved by the entire system. Taking the reciprocal of that divided by two will give you the spatial resolution of the system as a whole in lp/mm.
very nice explanation Jrista, and the first coherant technical epistle I've seen here regarding the effects of lens choice as regards the resolving power of the sensor, both the contribution of individual components and the asymptotic behavior of the function. Essentially, the 1/(2 * RMS) method suggests that when one component in the system is replaced by one that is significantly worse than the previous aggregate, that the effects will probabaly be noticed. Moreover, the effect of such a substitution will be more noticeable with there are fewer components in the system. Accordingly, using the approximation of only two components (the sensor/lpf and the lens), one can easily see that the choice of lens will influence the overall resolving power of the system. Captain obvious, to be sure, but one could model the equation and see the effects (on end-2-end resolving power) of choosing one lens over another, an excersize left "to the reader", lol. . I suspect most would rather look at photos though