When all we had to shoot with was film the limiting factor in terms of sharpness, besides the usual suspects like proper technique etc., was the glass we used.
With this I mean that, as long as you stuck to the lowest Iso, a better lens would have shown a better performance and, at a certain point, it would have over-powered the film capability of recording detail.
Finally we are getting there again with digital as well! Thanks to the latest advances in sensor technology – read: megapixels – we can at least be sure that glass will be the limiting factor in achieving sharpness, and not the sensor.
And if the latest 50Mp sensor of the Canon 5D series it seems too much to you, just know that the same Canon has just presented a new sensor – targeted mostly to video users and with a size between Aps-c and full-frame, though – of a staggering 120Mp. No, I have not added a zero by mistake: that’s really one hundredth and twenty megapixels!
You can check it here:
I don’t think that that many lenses will be able to take advantage of such resolution, but the point is that it will not be necessary. Keep in mind that the total amount of resolution of a system is not simply given by the sum of its components.
With this I mean that if you shoot with a lens capable of resolving 100Mp on a sensor of 100Mp you will NOT get 100Mp as a result, but more likely something about 65Mp or so. The math is something I will gladly avoid, and it has to do with the good Nyquist.
Think of it this way: the speed of a platoon is determined by its slowest member, and divide that speed by 1/3.*
So what happens when you have a sensor that resolves more that your lens is capable of? Right: you still get a better result, because the sensor higher resolution will “over-sample” the amount or resolution the lens is capable of, and less of it will be lost. Brilliant!
That being said, until they find something better than a Bayer pattern I think the sweet spot – to have better high Iso performance, as well as resolution – will be between 50 and 100Mp for a full-frame sensor.
But the moment they should begin exploring other techs, like Foveon-like layered sensors or organic sensors or whatever, then all the bets will be off.
Bottom line: we could in a few years find ourselves playing with a 500Mp organic sensor with 20 stops of dynamic range!**
*Please, you math guys and engineers just don’t eat me alive, I know I suck at math! **Yeah, I know this is mostly wishful thinking, but so it was shooting at anything more than 1600 Iso a few years ago.