Microcontrast

Started 4 months ago | Discussions thread
AiryDiscus Senior Member • Posts: 1,621
Re: Microcontrast

Joofa wrote:

AiryDiscus wrote:

Joofa wrote:

AiryDiscus wrote:

Joofa wrote:

fvdbergh2501 wrote:

Joofa wrote:

Great Bustard wrote:

  1. If System A has a higher MTF-50 than System B, does that mean that it will have greater resolution at all other contrast levels? For example, will the MTF-10 and MTF-90 of System A also be greater than System B?

IMHO, some of the answers you are getting in this thread are not entirely correct in the context that you have posed the question. (For e.g., the graphs shown here, and here.). You have posed the question in the context of a picture as in digital photography. The correlation structure in a natural image restricts frequency content to a range in the (relatively) lower frequency in an image. After that it is just noise. All that arguments in the quoted threads are actually trying to find (micro) contrast in noise! Even their own graphs seem to suggest that in the relatively lower frequency region (where most of the image content lies) the MTF curve which is higher stays more or less higher.

I think you misunderstood what my post shows:

I think that an issue that is not being stressed enough here is that I'm talking in terms of a photograph, and how in reality the typical corresponding frequency spectrum is distributed for an ensemble of natural images, and when such a spectrum passes through an optical system.

Both yours and Airy Discus' graph show that in low frequency region (below Nyquist) a higher MTF curve (of an optical system) stays higher. And, this is the range where most image content is. The image frequency content (in ensemble sense) goes as 1/f. Therefore, after treatment with an optical system of the type shown in the graphs linked, a higher graph still says higher in that range.

And, that answer GB's question in this range.

the OP wanted to know if halation (taken to mean spherical aberration) affects MTF50.

I didn't say anything about halation.

The MTF plot demonstrates quite clearly that:

a) The hypothetical SA-affected lens at f/1.4 only just manages to achieve a comparable MTF50 value, despite the fact that we would expect much higher MTF50 values in the absence of the simulated spherical aberration. This meets the first part of the OP's question.

b) The simulations also demonstrate that the contrast of a system suffering from spherical aberration can indeed be higher at higher frequencies, even in the case where the two systems were matched at MTF50. (second part of OP's question)

In other words, my post provided a simple counterexample to address one of the OP's questions. Nowhere did I claim that the additional contrast beyond Nyquist was a good thing; you must have missed the phrase "false detail (aliasing)" in the caption of the simulated image, or the later phrase "but in reality that is just aliased false detail".

I didn't say anything about contrast beyond Nyquist. In fact I had stuff below Nyquist implicitly without stating that word.

Regardless of the obvious aliasing beyond Nyquist, the simulated system with SA (green curve in the MTF curve plot here) does indeed have better contrast at high frequencies but still below Nyquist, which is consistent with the OP's definition of microcontrast.

At no point did my post claim that I supported the OP's definition of microcontrast, or that high contrast beyond Nyquist was a good thing.

Again, I'm not even talking about anything beyond Nyquist.

Riddle me this; if all of these high frequency details are unimportant/noise, why do we continue to see improvement to detail from higher resolution sensors and lenses for natural images?

Most image content still resides in a lower range of frequencies.

Of course, it's gone through a cascade of filters that attenuate higher frequencies more than lower ones.

Regardless of filters a natural image would have overwhelming large fraction of image content in the low frequency range. That is due to the statistical correlation property of natural images. Filtering can even emphasize that fact more, as you say.

Such a property cannot be general.  E.g. images of modern architecture (angular, many lines) will be dominated by higher frequency content.

And, no where did I say that higher resolution sensors are not important, in general. However, from the point of view of GB's question of so called 'microcontrast', it appears that you are concerned more about a range that has little image content to make a case regarding a higher MTF graph in lower frequency suddenly becoming lower than the other MTF curve in the higher frequency range.

No, the very first words in my comment were that I think it is the opposite end of the spectrum to GB (i.e. low frequencies).

But, then your graph shows that a higher MTF optical system stays higher in the low frequency region (irrespective of any aberrations that might switch it otherwise in the higher frequency region).

What?  I showed no more and no less than:

It is possible for two optical systems which have very similar MTF50 to have very different MTF at high spatial frequencies.

In another comment to you, I showed that:

It is possible for two optical systems which have very similar MTF50 to have very different MTF at low spatial frequencies.

It all depends on the content of the wavefront and other phenomena e.g. scatter.

So, GB question is answered based upon your graph: In the low frequency range, where most image content resides, a higher MTF optical system remains higher everywhere in this range.

No, such generalizations cannot be made.  bob2n even showed measurements (real data!) from Zeiss showing this.

Post (hide subjects) Posted by
(unknown member)
(unknown member)
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow