MattGrum

​MTF charts published by lens producers are already greatly reduced in complexity, in that in general they only show contrast for two spatial frequencies (usually 10 and 40 lp/mm) and two aperture values (usually wide open and f/8) and two image plane orientations (sagital and tangential). The "full" MTF is a four dimensional quantity which is difficult to represent in 2 dimensions. I put "full" in quotes because it also depends on focal length for a zoom lens, and, to an extent, on the focus distance thus it can be a six-dimensional object. I hope you can appreciate why it's impossible to boil a six-dimensional function down to a single number. You could simplify it further and measure the MTF at a few distances from the image centre (centre, top/bottom edge, left/right edge and extreme corner), but it's been heavily simplified already and that would cut a lot out. You could try averaging the figures but the range you average over would be fairly arbitrary and subject to differences between manufacturers. And the idea that manufacturers attach a resolution figure to the lens in order to sell them would just open the flood gates to clever ways to inflate the score by carefully choosing what to include/exclude.

No they really shouldn't. A lens does not "resolve 6MP". A lens renders a certain spatial frequency at a certain contrast ratio. It might take detail at 20 line pairs per millimeter and produce 70% contrast, detail at 50 line pairs per millimeter results in 30% contrast etc. This relationship is captured by the modulation transfer function (MTF), a quantity which varies according to the distance from the image centre, and the direction you measure in (sagital vs tangential). Manufacturers already publish MTF charts for their lenses, which is the equivalent of what you're suggesting, only much more meaningful. There are some differences in how these charts are computed (e.g. whether diffraction is included or not) so they're not always directly comparable, but they aren't anywhere near as misleading as trying to attach a single "megapixel" rating to lenses. The other reason stating "this lens resolves 6MP" is meaningless is that the important thing in determining how you images will look is not the lens MTF itself, but the system MTF. The system MTF is the product of MTFs of each part, the lens the filter stack and the sensor (and image processing to an extent). Because it's a mathematical product (a lens delivering 80% contrast combined with an AA filter that delivers 95% contrast results in 76% contrast (0.8 x 0.95 x 100)) you can improve the system MTF by improving the MTF of any component in the system. Hence you "6MP" lens will give you more resolution on a 24MP body than on a 6MP body. It's exactly this thinking that leads people to declare that there's no reason to have a 50MP sensor as there are no 50MP lenses in existence. Even the kit lens in your example produces some contrast in the centre of the image at 50MP.

​ Firstly lenses don't "resolve" a certain number of megapixels, what happens is contrast starts to go down as details get closer together. This is characterised by the MTF of the lens. The sensor has it's own MTF. System resolution is determined by the product of the lens MTF and sensor MTF, thus an improvement in either can lead to an improvement in system resolution. Secondly most EF lenses will provide excellent results with a 50MP sensor over the most important central region of the image, as demonstrated by years of FF lens use on APS-C. I'd bet many would perform adequately even at the borders. They certainly wont be any worse than on a 22MP sensor, whilst at the same time you will get less alisaing on demosaicing artifacts by virtue of oversampling. It is sad to see the artificial segregation of the camera through downgrading video features and ISO settings. I'm very much hoping Sony can manage a line-skipped 1.1x crop 4K video from their next 50MP sensor, that should resize down to a very nice 1080p file.