Why micro-contrast is hard to measure

Started 1 month ago | Discussions thread
Flat view
fferreres Senior Member • Posts: 1,577
Why micro-contrast is hard to measure

Microcontrast is contrast at higher resolutions or the lens ability to preserve fine detail. A lens performance can be measured doing an MTF test. Taking these tests isn't easy, and especially so since actual sensors have filter stacks in them that depending on many factors, can affect what you actually get in real pictures.

It's hard to measure

But assuming it is the same, the typical (majority, not all) user will not see MTF charts, or if he/she sees it, will not gain much from them. But even it sees them and understands them, and tries to compare a lens, these may not even be compatible at all, as you'd need to compare at the same f-stop, and the methods to compute this MTF charts would also be the same, and the lenses used for testing purposes may have been especially pre-tested and may not be what you get (especially, for manufacturers that have lower quality control and precision in the build process across everything that can affect final outcome). And even if it finds these MTF charts, they'd need to be representing the same. Both should be real and not theoretical.

Both should be done with white light, and not green or limited wavelengths. And even if these are the same, and same f-stop, both should also be rated in the same scale of lp / mm. For example, most lenses are rated on 10 lp / mm and then 30 lp / mm, where as Zeiss lenses usually show 10 lp / mm, 20 and 40. But even if you are to be able to test this, they will usually only show the average of all wavelengths, where a more useful chart will show not just tangential and saggital, but should show for example, across what the an RGB sensor would, across red, blue and green wavelengths. And even then, this will not be enough to really compare them because most lenses tested will chart the MTF charts with lp / mm in Y and then distance from the center in X. This gives a sense of how it resolves line pairs as you move off center, usually either tangential or saggital would suffer a lot more.

But in reality, if this would be ok, this would not at all be enough, because you can't see how this evolves over time (what happens in between 12 lp / mm and 28 lp /mm? Some may hold at 12, some may abruptly fall to 28 lp / mm). You'd need to map how it transitions as we further and further increase lp / mm.

But if you had this full transition (now X is lp / mm) that's not enough, you want to have Y axis showing contrast. A lens can't create more contrast than is there in the source (except special immensely unimportant unnatural cases) and can only reduce it through aberrations or lowered transmission. So you'd want to actually see the comparison using the same testing equipment in real life, at least for a generic case with no sensor stack (but then again, with any particular sensor it might be different). So you'd want to see in the Y axis how much white light contrast at an ever increasing density of line pairs (lp /mm). A chart lie this.


If you had something like this, for different f-stops, and one for each lens, and for each lens and f-stop, at different distances from the center up to corners, then, you'd be able to maybe talk about micro contrast of the lens scientifically. But it would have to be the lens you own. if it's some other copy, and the manufacturer doesn't have extremely tight manufacturing tolerances and then test every single lens to ensure it's within, then, you will not have what was tested. You'd only have something else, in particular, you'd have what you actually got, which would likely be different (and worst).

Once you see these, you'd have to note that pixel peeking would affect which lens has "more" microcontrast only with regards to a particular resolution. A lens may have a quick dive in lp/mm at 30 microns, and stay relatively stable there after, and one may have a progressive decline where the two MTFs cross, so one lens, if you consider certain resolution, may have a lot better microcontrast (it still hasn't reached the pint where the lens resolution crumbles or takes a deep dive) but then at 50 lp / mm, the lens that LOST at 30 lp / mm may now be way above the winner at 50 lp / mm (because it got worst much more slowly than the other after 30 lp / mm).

So talking about microcontrast of lenses, without any MFT mapping % contrast with lp / mm at certain f-stop using the same methodology and the same lens, and assuming you'd get equally performing lenses if you order, and given how we use different cameras, with different sensor stacks, and with different resolutions, and formats (sensor sizes), etc. and that we may compare this at different levels of downsampling, and also apply a lot of additional processing, makes all these discussions nonsense.

And even then, this is not really the full story. I don't want to make this post 76 pages. So to name one example...how does the lens perform in backlist situations? The MTF did not test how your lens would handle flares. How much light not within the FOV is distributed? How? How does a very strong light within FOV affect the image? Again, you add up all that was stated before, and still can't reasonably say that one has more microcontrast. It will depend on all the before, but also how it handles flares. And then, you can start adding other factors as well...never ending.


Now back to resolution (or pitch size), just to give an idea of how much when we pixel peek, we aren't even aided by any normal manufacturer provided MTF that is published in marketing literature.

Max Lines per mm for different sensors:

A7II sensor: 83 lp / mm when you pixel peek

A7r IV (unreleased): 136 lp / mm when you pixel peek

EM1 Mark 2: 148 lp / mm when you pixel peek

A lens that can effectively retain across all 3 color wave bands (relatively) high contrast at the pixel level is obviously very very different for a user of A7II, future A7r IV and EM1 Mark 2 user.

This also of course varies a lot depending on the F-stop. As more resolution is added (actually, as pixel becomes smaller in microns) the aperture starts playing a big role. Even at f5.6 you will may start to be diffraction limited in terms of ability to retain contrast, as the airy disk may already be a bit larger than the pixel size. So your lens that is "very sharp" at f11, is already not too sharp at these resolutions by definition. if it needs to be stopped down to become sharp, don't expect a lot of contrast at high resolution.

A lens with high micro contrast, ultimately, will depend on your target resolution, the real MTF chart of the thing you have in your hand (which will vary for all parts of the sensor surface), the colors prevalent in the photo and your lens treatment of each band, the sensor stack, the lighting conditions, the actual control of all aberrations that can affect contrast, the quality of any filter you use, etc. etc. etc.

People with lots of discerning ability and experience can detect it

Now, finally, people that have used many lenses, ideally, a lot of expensive, cheap, different designs, and shoot a lot, and are demonstrably extremely discerning and picky, WILL tend to note or if not note, gain a real sense of which lenses they prefer. They have used many lenses, they are not fanboys of anything (the ones I am talking about), have used different resolution cameras, etc. These people have not seen all these very detailed MTFs, but may have peaker at 70,000 pictures over 150 high quality lenses, over and over and over. And they will retain some lenses. And if they are obsessed one bit, they have a good enough sample for them to tell (for some resolution, and in their experience) which lenses where "more crisp", had more "punch", or they believe have more "pop" (which would add bokeh and other factors into the mix).

For example, Fan Boy Lloyd Chambers expresses his opinion, but focuses mostly on Zeiss lenses (he says nothing about other brands as it's a Zeiss blog). Of course, no amount of 10,000 MTF charts (and these are nowhere to be found either) will replace his experience in having used so many lenses, in so many scenarios, to know what he is talking about. Whether you want to believe him, or someone else, is different.


But however you believe, or even if you don't believe anyone, the higher you go in resolution, yes, unfortunately, micro contrast is about having an awesome resolving high quality sharp lens as wide open as possible, in a way that is hard to measure (re read entire post)...as the article above sums up, in a sense "it's the sum of everything" (al least with regards to resolution/contrast levels).

Final words

If you don't see it, or don't believe in it, you are lucky! The only ones that are affected are those that have used a lot of lenses and have become a little obsessed by the effects. Count me as one, but not so much obsessed, but I have had a good number of opportunities to compared about a hundred different lenses. I see it. And it makes life more complicated!

Olympus E-M1 Sony a7R
If you believe there are incorrect tags, please send us this post using our feedback form.
Flat view
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow