Restating the Q formula from this post with the Bayer CFA correction from yesterday’s post:
Qbayer = lambda * N / (pitch *1.7)
Where lambda is the wavelength of the light in micrometers, N is the f-stop, and pitch is the pixel pitch in micrometers.
For 0.5 micrometer light,
Qbayer = N / (pitch *3.4)
Since a Q of 2 means that the sensor and the lens are “balanced”, we can relate f-stop and pixel pitch for balanced systems:
N = 6.8 * pitch
pitch = N / 6.8
For a setting of f/6.8, we want a 1 micrometer pixel pitch. For a setting of f/8, the pixel pitch should be 1.18 micrometers. These numbers are much finer than any available sensors sized at micro 4/3 and larger.
If we think the correction factor should be one,
Q = lamda * N / pitch = N / (pitch * 2)
And for a “balanced” system
N = 4 * pitch
Pitch = N /4
At f/8 we want a 2-micrometer pixel pitch, still finer than currently available for any available sensors sized at micro 4/3 and larger.
The bottom line is that any interpretation of applying image system Q and the idea of the balanced system to Bayer arrays gives the result that, for fine lenses, resolutions of at least a binary order of magnitude higher than currently available are desirable. This argues for 200+ megapixel full frame sensors.
This is at variance with conventional wisdom.
Erik Kaffehr says
I don’t think lenses are re really diffraction limited from the Imatest MTF measurements I have down. Clearly, diffraction affects MTF but MTF is quite a bit below what would be possible, So I would suggest it is the combined effects of residual aberrations and diffraction we see.
Could be interesting to check out that 41 MPixel Nokia phone, may be an interesting example of what is realistically possible.