The folks at DxOMark have a new measurement, Perceptual Megapixel. This purports to boil down the information from modulation transfer function (MTF) graphs into a number. It proves that a camera with an enormous number of pixels isn’t all that useful unless you have an amazingly high quality lens and, probably, have locked the camera/lens down to a tripod. Folks who have a 20 megapixel camera and a $200 kit lens might be better off simply capturing at 6-10 megapixels. See, for example, a test of a Nikon superzoom lens in which 6 perceptual megapixels was all that could be extracted (third party superzooms came in at 4 or 5 megapixels). This is the resolution that Kodak selected for its consumer PhotoCD system back in the early 1990s.
Summary: the electrical engineers have pushed sensor resolution far beyond the capability of any optics that ordinary consumers are willing to purchase and carry.
[Separately, one of the first tests done using this metric shows that the $900 Sigma 35/1.4 lens dramatically outperforms both Canon’s $1300 equivalent and an $1800+ Zeiss manual focus lens. (The failure of prestige names to dominate objective tests is not new. I remember years ago a European photo magazine did optical bench tests of 50mm lenses and concluded that the $100 to $200 Canon and Nikon 50/1.8 lenses outperformed Zeiss and Leica lenses costing up to $2000. If memory serves, the magazine selected the Nikon 50/1.8 as the best lens overall, considering the balance of sharpness and distortion.)]
On the technical front, that’s handy. The consumer market itself is finally moving on from megapixel madness (to other fads, the most recent being “bokeh” ability, which is really upon the lens). My question is of the relative merits between all things marketers can seize upon. I never thought much of pixel counts, but do think more of sensor size (e.g., RX100’s 1″; XF1’s 2/3″) though I suppose that could be gamed as well.
With the superzooms (the one you cited has a 17x zoom range!), you expect — or should expect — that you’re trading off quality for convenience. They also generally have a lot of distortion and vignetting, though those can often be automatically fixed quite well with lens profiles in tools like Lightroom. Very convenient for light travel.
About the “perceptual megapixels”, it obviously feels easy to interpret the number, but I wonder how accurate that obvious interpretation truly is, and what generalizations we can derive and interpolate from what is a gross generalization to begin with. It seems to be a really convenient way to compare lenses, and to understand the appropriateness of a particular lens for a particular body, but it’s probably a good idea to keep in the back on one’s mind that there’s some hand waving we don’t truly understand.