Photo examples of sensors outresolving lenses?

Euell wrote:

JimKasson wrote:

Euell wrote:

JimKasson wrote:

Euell wrote:

JimKasson wrote:

Euell wrote:

Bruce Oudekerk wrote:

I strongly suspect that any lens that performs well on the other Rs will be satisfactory on the RIV. It's not really a huge jump in resolution. Good lenses tend to resolve more on higher MP sensors and really good lenses really take advantage of that extra resolution. Its also true that undesirable traits tend to get more obvious on higher resolving sensors. At some point good lenses transition into unacceptable lenses but I'm not convinced 'out-resolving' is the term of choice here.

Bruce

Let's take an example where a well regarded lens performs worse on a higher resolution sensor.

If the physical size of the sensors are the same, a given lens will perform the same or better on a higher-resolution sensor than a lower-resolution one, if the comparisons are made as same image size.

Jim

So, Jim, how would you explain DxOmark's resolution finding that most lenses perform better on the sensor with lower resolution in the case of the Sony A6000 and Sony A5000?

I never try to unpack DxO's results. I have not been able to reproduce them, and some details of their methodology remain opaque. Are you sure you're not talking about pixel-level comparisons?

BTW, I just posted a comparison of the same lens on a 50 MP and on a 100 MP sensor. Both pixel apertures are about the same.

https://blog.kasson.com/gfx-50s/fuji-100-200-4-on-gfx-nikon-70-200-2-8e-apo-sonnar-135-on-z7/

You may find the differences interesting.

Jim, why don't you take a look at what DxOmark has posted, including their "measurements?" If DxOmark is correct, then the pixel pitch must exceed most lenses' resolving power.

I have looked many times, never figured out the fine details, and I trust my own measurements.

You do realize that what you are positing, if I understand it correctly, contradicts the mathematics of digital sampling theory, don't you?

Please, do explain. Would DxOmark contradict the mathematics of digital sampling theory? I do note, however, that sometimes mathematics comes up with reason-defying results. For example, the Monte Hall problem in which there are prizes behind three doors. The contestant guesses one and Monte opens another. Should the contestant change his or her original guess. The answer is yes and can be proven mathematically, or so it is said.

There is one right answer to that question, but it's counterintuitive. That problem wouldn't slow a real mathematician down.

OK back to image processing theory. To correctly reconstruct an image, it must be sampled at at least twice the highest spatial frequency contained within the image. Sampling it at a lower spatial frequency will result in ambiguity which will cause artifacts in the reconstructed image. Making the Pitch finer raises the sampling frequency. Making the pitch coarser lowers the sampling frequency.

In no case, even with all spatial frequencies in the image less than half the sampling frequency, will reducing the sampling frequency make the reconstructed image more accurate.

Tip of the hat to Harry Nyquist, who figured this out in 1D.

https://en.wikipedia.org/wiki/Harry_Nyquist

Jim