Jonny Boyd wrote:

I think part of the problem in the previous thread is people talking about different things using the same terms, or the same thing using different terms, or disagreeing about which tiny mathematical improvements should be regarded as real world differences, and which should be viewed as negligible changes. Rather than rehash the existing discussion, I'd like to approach things from a slightly different tack.

If we take the system resolution equation r = (l^-1/2 + s^-1/2)^-1/2 where r is the system resolution, l i the lens resolution, and s is the sensor resolution, then we can draw the following conclusions:

As l tends towards infinity, r approaches the limit s, as s tends towards infinity, l approaches the limit r, and when l and s are roughly equal r = l * 2^-1/2 = s * 2^-1/2.

In other words, as you improve sensor resolution, the overall system resolution increases, but can never be greater than the lens resolution. Similarly, as lens resolution improves, overall system resolution improves, but can never exceed the sensor resolution. In other words, there are hard limits on the resolution of the system.

Now imagine a system with resolution r_1 featuring a lens with resolution l_0, and a sensor with resolution s_1 = l_0 * 10^1/2. Also imagine a system with resolution r_2, featuring the same lens, but a different sensor with resolution s_2 = l_0 * 10^-1/2. Therefore s1 = 10 * s_2

r_1 = l_0 * (10/11)^1/2

r_2 = l_0 * (1/11)^1/2 = s_2 * (10/11)^1/2

The resolution of r_1 is effectively equal to the resolution of the lens, while the resolution of r_2 is effectively equal to the resolution of the sensor s_2.

Increasing the resolution of the sensor by an order of magnitude results in a swing from system resolution being determined predominantly by the resolution of the sensor to the system resolution being determined primarily by the lens.

Both systems should have their peak sharpness at about the same aperture, but the limiting factor in the resolution will be different for the two systems. Increase sensor resolution further beyond s_1 and won't get a high resolution than l_0 so it's effectively diffraction limited. Decrease the resolution below s_2 and it's predominantly the sensor resolution that determines the system resolution so diffraction is largely irrelevant. Hence some of us saying that r_1 is more limited by diffraction than r_2 even if peak sharpness is at the same aperture and r_1 is always greater than r_2.

Glad to see that you finally worked out the implications of the formula I gave you.