Microcontrast - real or virtual?

I read a lot on the net about ‘microcontrast’ with some swearing by certain lenses that display it, over others that don’t. As far as I can see, it can be explained by lens resolution and contrast, but where does the ‘micro’ part come in, if indeed it does?

"Microcontrast" is one of the unicorns of photography. It does not exist. Take this cum grano salis because I don't really believe in "bokeh" either (it's really blur).

Maybe some people use it to refer to something that doesn't exist, but some people use it to refer to something that does. For an explanation, see this article.It's used to refer to the ability of the lens to resolve differences between small details. This is very different from the usual use of contrast, which refers to tonal values. For example, an image that extends from nearly fully black to nearly fully white has more contrast than one that is mostly mid-tones, and among two images with the same tonal range, the one that has greater spread in the midtones will be perceived as higher contrast (hence the S-curve using a curves tool). The resolving ability of a lens has nothing to do with the range of tonal values.

The more I read the less I seem to understand. There seems to be a consensus among many, at least, that there is such a thing, but even those who agree it exists differ in their definitions, and there's less certainty about how it's achieved when it is, or whether it's visible when it occurs. In particular, the question of how micro contrast relates to resolution seems quite variable.

Micro contrast is a term that crept into use with its co-conspirator 'acutance' during the 1960s. It was first used in reference to the (elusive) edge-effect seen in B&W film with stand development, or with developers specially designed to give the effect.

 

Having used a few such developers, IMHO 'acutance' developers are a load of hooey. (Google 'Mackie lines' for a fuller explanation.)

 

Then along came scanning and digital cameras, where USM or other sharpening techniques enhance the edge-effect electronically. This really does increase micro-contrast without affecting overall contrast, giving the illusion of greater detail.

 

So, yes, micro contrast does exist. It simply describes what happens to an MTF curve at higher spatial frequencies. It's a confusing term, because it implies there's some hard cutoff point where normal contrast becomes 'micro' - not so.

Edited August 2, 2018 by rodeo_joe|1

I agree with Rodeo Joe. The history and the rest too.

 

BTW I regard (and I use) the meaning and the use of term "acutance" as a measure (often by an A/B comparison) of sharpness (edges under close-up scrutiny), as opposed to a subjective evaluation of the print at a viewing distance.

 

'micro contrast' I align more with the results of the Post Production of a Digital File, rather than as a feature of a lens: (the way I interpret Rodeos Joe's commentary, he was saying that, I think.)

 

WW

"(the way I interpret Rodeos Joe's commentary, he was saying that, I think.)"

 

- Ermmm, not quite. Micro contrast existed, both in reality and as a concept, long before digital manipulation became common. It just refers to the way that a lens or film or sensor, or combination of above, handles fine detail.

 

Technically it refers to the high frequency (more cycles or line-pairs-per-millimetre) end of a resolution chart or MTF curve.

 

Confusion sets in, I think, because the word 'contrast'* is used too sloppily, and there's no agreed point at which plain old contrast becomes micro-contrast. And even if there was, that definition would fall apart at large print or display magnifications. Hence the supposed poor performance of perfectly acceptable lenses under pixel-peeping conditions.

 

*Contrast: How tones in a subject are rendered in an image of that subject. I.e. how deep or light a shade of image tone is given by its corresponding subject tone. This applies equally to the black/white bars of a resolution chart, the sinusoidal variation of brightness of an MTF target, or the highlights and wrinkles of granny's face.

 

I'm all in favour of going back to actually printing our photographs, and hanging them on real walls instead of in virtual galleries. Maybe then we could forget most of this technical BS and concentrate on lighting, composition, mood, subject interest and all the rest of the stuff that really matters.

"(the way I interpret Rodeos Joe's commentary, he was saying that, I think.)"

 

- Ermmm, not quite. Micro contrast existed, both in reality and as a concept, long before digital manipulation became common. It just refers to the way that a lens or film or sensor, or combination of above, handles fine detail. . . [and to the end]

 

OK. Thanks for the detailed explanation/expansion. I was unsure exactly what you meant. That's clarified it for me.

 

I was ignorant of the fact stated in the sentence underlined and bold. Thanks for that and it forms much of the clarification for me: I'll research that particular subject later, for my own interest.

 

FWIW, I am not that much into screen based technical "BS", as you put it. I like prints, hung on walls.

 

WW

 

PS - One can learn much, reading the "Beginner" Forum!

Flare is present in all optical systems. Flare is caused by stray light rays that reverberate about in the optical system. Some are rays from bright sources just outside the field of view of the camera. Some are image forming rays that hit the blades of the circular aperture. Some are rays that hit shiny object inside the camera body and are reflected. Some are uncorrected residual lens aberrations. In any event, these stray rays eventually make their way and bathe the image area. The result is flare and flare is devastating. Flare reduces contrast. The chief mitigators of flare are coated lenses, flat black camera interiors, and the use of a lens hood to shield the camera from rays coming from outside the principal subject area.

Thanks for all the responses. It obviously can’t be easily defined, which is reason enough for me to file it away in the ‘Mostly BS’ file.

wogears, "Microcontrast" is one of the unicorns of photography. It does not exist. Take this cum grano salis because I don't really believe in "bokeh" either (it's really blur).

 

The word origin of bokeh comes from the Japanese word ボケ which should be read as bo-ke. The word translates as blur, haze.

wogears, "Microcontrast" is one of the unicorns of photography. It does not exist. Take this cum grano salis because I don't really believe in "bokeh" either (it's really blur).

 

The word origin of bokeh comes from the Japanese word ボケ which should be read as bo-ke. The word translates as blur, haze.

 

Thanks for that James. I never realised the Japanese word translated simply to blur or haze. I thought it was something much more esoteric and arty-farty. Now we know..... the extent of pretension attached to both the word, and associated pointless examination of out-of-focus highlights that in most cases would have been better excluded from the scene.

Thanks for that James. I never realised the Japanese word translated simply to blur or haze. I thought it was something much more esoteric and arty-farty. Now we know..... the extent of pretension attached to both the word, and associated pointless examination of out-of-focus highlights that in most cases would have been better excluded from the scene.

 

I'm guilty of talking about the bokeh as a characteristic of the lens...because let's face it-especially with large aperture lenses-it's one of the things that really defines the character of the lens. Although the highlights are certainly part of it, I consider the whole look of the OOF area.

 

With that said, I've seen/heard a lot of folks talk about taking "bokeh photos" where they mean anything where the background is out of focus.

I suspect that bokeh and microcontrast share with many other things the distinction of being real, but less important than some people would like. We can always find some examples of lenses in which the out of focus rendition is so lousy that it hurts, and some in which it's great, and probably could also find a few lenses whose microcontrast is awful, where some are splendid. But the further from the fringes you get, the smaller the difference, and the less important to the overall image. I think some people get distracted by the nearly invisible differences that you are unlikely to see unless you're looking for them.

I think some people get distracted by the nearly invisible differences that you are unlikely to see unless you're looking for them.

It's hard to argue with a magnificently composed and perfectly exposed image of an engaging subject. f/8 and "be there" still rules over all the esoterica.

Wait... it's a bit hard to define in exact terms, so the concept must be rubbish? I turn my back on a forum for five minutes... different countries have different definitions for the age of consent, or for driving, or for buying alcohol - that doesn't mean there shouldn't be one.

 

I'll take a stab, with some confidence that I'm in the right ball park but the disclaimer that someone could tell me I'm talking rubbish.

 

Let's start with a digital image (because I'm a computer person), and we'll move to lenses afterwards.

 

Imagine you have a 6 megapixel image (call it 3000x2000) from your camera. It contains detail at every pixel, but only six million pixel's worth of information. If you blow up that image to 24 megapixels (6000x4000), the new pixels are generated by interpolating/averaging the ones you've already got. In terms of storage, the resolution of the new image is 24 megapixels, but in terms of the amount of image data there, it's still only six. If your 6MP image had a photo of the roof of a house, but you couldn't make out the individual roof tiles, you wouldn't be able to make them out just because you'd rescaled the data to more pixels - but you might be able to make them out if an actual 24MP photo was taken of the same roof. There is an absolute limit in a digital sensor on the amount of detail you can see.

 

Now, disclaimer, there are things you can do to get more information into your image. You can get extra detail from taking more photos. You can use an algorithm that detects what seem to be edges, or seem to be repeats of other areas of the image, or you can use a neural net and train it on what detailed areas of images that look like yours should look like. But you can't guarantee an exact match to reality based just on the pixels you captured in the original image - at some point, the algorithm is making up data that doesn't exist, and it might get it wrong. This is why software that enlarges images looks bad once you've gone too far. From a mathematical perspective, the only thing you can safely do is blur between the pixels you've got - then you're not introducing new and possibly spurious data.

 

The contrast (not micro-) of an image is the difference between its darkest black and brightest white. The more contrast the punchier an image looks, and the easier it is to make out detail - but the harder that detail is to capture. Some lenses have "low contrast" because light bounces around inside them, and this lightens the darker areas of the image.

 

Microcontrast refers to how much light spreads out in a small area when the lens tries to focus on it. There are no perfect edges - point a lens at a bright light and the area of image around that light will also get brighter, a bit. How much area and how much is a matter of microcontrast. I'll come back to this.

 

Consider our digital image. We can run a sharpening algorithm on it: what these do (indirectly) is detect an edge between brighter and darker areas, and make the brighter areas brighter and darker areas darker. This has similarities to adding more detail - if you have a high amount of detail in an area you might expect adjacent pixels to vary a lot in colour, and if there wasn't much detail you might expect there not to be much difference. But sharpening adds microcontrast, not additional detail. No matter how much you sharpen the roof in our up-scaled 24MP image, you're not going to start seeing tile details that weren't there in the first place.

 

Lenses aren't like digital images. The (micro-)contrast with which they represent detail drops off as the detail gets finer. If you take a photograph of a chess board from close up, the white and dark squares will be pretty white and pretty dark. Move away so they're small iln the image and they'll start to blur together, until eventually you just get the whole board looking the average colour of its squares. If you want to shrink a digital image and get it to look right, you actually have to emulate this behaviour in software.

 

Lenses often have published MTF charts, both of which plot measurements of the lens at distances from the centre along the horizontal axis. There are two forms of these. In the first, there's an arbitrary "line pairs per mm" value - usually two on the same chart - which shows how much microcontrast the lens exhibits when putting that much detail on the sensor. Since the numbers are usually a bit different if you're measuring along a line away from the lens centre or in a circle around the lens centre, you often get two values plotted at each LP/mm value. At a LP/mm (less resolution), microcontrast will be higher than at high LP/mm (fine detail). Good lenses retain almost perfect detail at a low LP/mm value (the graph is roughly a horizontal line at the top of the chart). Very good lenses nearly match this with a higher LP/mm value.

 

But this just tells you about the microcontrast at those two arbitrary levels of detail. What if your camera can resolve a lot more detail than that? What if it can only record less? The other kind of MTF chart shows the LP/mm on the vertical axis, and plots the level of detail at which the lens has a microcontrast above a specified amount (such as 50%). To an extent, that tells you better how well a lens might hold up to your camera. But that "50%" is also a bit arbitrary. It's be lovely if people plotted more 3D graphs that gave you a better way to check microcontrast at every level of detail - but they don't.

 

Generally, a lens with a lot of microcontrast (at a given LP/mm) will show detail at that level better than one with less microcontrast. On film, which has a similar gradual fall-off of detail against contrast, this is quite important. But it's not quite the same thing as resolution - a lens may be very contrasty at one level of detail, but degrade more quickly beyond that point than another lens. Digital sensors record up to an exact resolution point but no further. Within the detail you capture, you can get some microcontrast back by sharpening the image in software - but this introduces noise, and can introduce fake edges. Still, lens correction software does some of this.

 

Bokeh is the quality (not amount) of an out of focus region. "Good" bokeh has a few features. As you move away from the centre of a blur circle, it's monotonically decreasing (that is, the light falls off with distance but never gets greater again). For most lenses, there's a large area of fairly constant intensity, then a smooth roll-off. Zooms, particularly with aspherical lenses, tend to have "onion ring" bokeh, which has concentric brighter and darker circles in the blur area. This looks distracting (the assumption is that you'r;e trying to make the out of focus regions of the image smooth, and "fake detail" like this can be distracting. Some lenses have a "rolled condom" bokeh look, where there's a brighter outline around a fairly smooth region. If you sharpen an image, this tends to be what you get (one of the things that gets sharpened is the edge of a blur circle) - so a lens with good bokeh can be made worse by processing if it has poor microcontrast. With the right subject, bad bokeh can be very distracting. Some lenses have a very smooth fall-off of light (often due to an extra masking element inside the lens) - they tend to capture less light in total, but the background is very smooth.

 

Other aspects of bokeh include its shape (round tends to be less distracting, and flat aperture blades give you polygonal bokeh) and any colour fringing it may have. Some lenses deliberately don't focus perfectly, meaning that there's a region of depth of field that's in pretty much constant (but imperfect) focus before smoothly transitioning away; other lenses can be a bit distracting in the transition from sharpness to blur.

 

So: Microcontrast exists, it's just not consistently plotted. Resolution exists, and it's tied with microcontrast but not quite the same thing. You'll see both of them if you peer closely at an image, but they're not the only aspect of image quality - and non-micro contrast also has an effect. Bokeh definitely exists, and you can see it from a distance in some images, because the blur circles can be large. However, it depends on the scene - if the background was already smooth, you won't see any roughness in the lens's rendering of it, but if you photograph a set of out-of-focus Christmas tree lights, bokeh is pretty much all you'll see. Most images are between the two.

 

I hope that helps, and I've not been too misleading? None of this is as important as taking the right photo of the right thing at the right time, or of getting the exposure right. But it's not irrelevant if you want the best overall image quality.

Andrew, the main difference between a simple resolution chart and an MTF curve, is that the lppmm chart is generally a printed target that has a limited brightness range and only pure 'black' or 'white' bars at specific spacings. The brightness capacity of a sensor or film may well exceed the limited range of such a chart, limiting and falsifying the resolution limit. OTOH, an MTF measurement contrives to present the lens or sensor with a sinusoidal variation of brightness over a continuously varying range of spatial frequencies. The brightness range of which is made to equal or slightly exceed the film/sensor saturation level. Therefore a true 100% contrast can be achieved at low spatial frequencies, and at higher frequencies a more accurate assessment of delta-contrast and eventual cutoff (0% contrast) frequency can be made.

 

I'm afraid you've been guilty of quoting one of those muddled descriptions of 'contrast' that lead to confusion. It's not just a question of how dark or bright some areas of the final image are. It's how those areas relate to the original subject brightness. And of course that's also dependent on the size of said areas, the size of adjacent areas of differing brightness, and the degree of brightness variation. Those are the many variables that an MTF curve attempts to show, and that a simple resolution target fails to address fully.

 

In practise, if you're not liking the micro-contrast of a particular lens, just stop it down!

 

I do agree that the OOF rendering of some lenses is quite ugly, but there are optical compromises involved in balancing spherical aberration (the main culprit for good or bad 'bokeh') with other nasties like LoCA or focus shift on stopping down. Personally I'd rather see a bit of hard-edged OOF blur than having coloured fringes around it.

Okay, so "macro contrast" (the extent to which incoming light is spread fairly uniformly throughout the image) is a thing, but hopefully a small thing in a modern lens unless you're actually picking up flare from the sun. Modern coatings improve overall contrast by reducing internal reflections within the lens; you can get lightly diffuse screw-in filters designed to reduce scene contrast to the extent that a camera can capture shadow detail properly while retaining highlights.

 

"Microcontrast" is a local feature, and the extent to which light is spread to the surrounding area, a function of the optical design (and less so internal coatings). I did talk about a sensor, in that I meant the output of the lens in the image plane, which is the effect on the image of a lens's microcontrast behaviour - and I concede that the technical definition of MTF (rather than my "guide to interpretation") relates to signal processing and the frequency domain. But, er, beginner forum. :-) I'm not sure where the talk of a resolution target came from, but sorry if I caused confusion there.

 

Still, full MTF measurements would tell you, for each distance from the lens centre and each of sagittal and meridional axes, the microcontrast at each frequency - defining, for the lens at a given aperture, a set of two-dimensional planes in three-dimensional space. The usual graphs presented show either a cross section or a contour through that plane. I just think it would be nice in this day and age of computer informatics to show the whole thing.

 

I switched to Nikon in part because smooth bokeh (visible at larger scales than absolute sharpness) matters to me, and I believed in magical tales of Nikon's "defocus control" lenses. I since learnt that they achieve smooth bokeh at the cost of colour fringes, which is not a trade off I'd have made knowingly. Fortunately newer optical designs (often with more elements, aspherical shapes and different glass formulae with different refractive indices) are able to achieve a good rendering and sharpness with fewer compromises - the most obvious issue with several recent lenses being "cat's eye" mechanical vignetting of bokeh.

 

We live in good times, optically, except when we have to pay for new glass!

Sure it's a beginner forum, and that should mean we provide clear and accurate information - to the best of our ability. I don't think it necessarily means we should simplify the issue or patronise the OP by underestimating their comprehension.

 

I still think that a description of 'contrast' that links input (subject) brightness to output (image) density/brightness is applicable to any part of an image, big or small, and doesn't need to be split into macro or micro compartments. Contrast is just contrast. The fact that it varies with subject or lighting conditions, with spatial frequency and across image zones really doesn't change that basic definition.

 

It's a problem of semantics and not optics.

I’m a simple man. I only understand simple things. Thanks so much for taking the time in responding. The responses have all been very informative and educational. Out of curiosity, I was trying to establish whether microcontrst is actually something that can be reasonably easily defined and understood, has some consensus of definition, or, alternatively, is a more ‘baggy’ concept that invites some variety of opinion. Thanks for giving me the answer I sought.

Joe: I do think there's a distinction between contrast at the whole image level caused by internal reflections and that at the micro level, as a consequence of lens aberrations. The latter is certainly a continuum with no clear upper limit to "micro". My 70-200 has pretty good microcontrast by most measures, but I significantly reduced the overall image contrast accidentally recently by getting tissue fluff on the front element (I'd stored the lens cap in a pocket with a tissue and not noticed the lint). Overall contrast isn't really an issue with modern lenses used properly, I admit.

 

Stuart: I'm glad if we helped, and sorry if I caused more confusion. Microcontrast is a thing, and although parts of the way measurements are presented can be a bit arbitrary, it is measurable. But it's also just one aspect of lens design, and often misdescribed.

(snip)

Consider our digital image. We can run a sharpening algorithm on it: what these do (indirectly) is detect an edge between brighter and darker areas, and make the brighter areas brighter and darker areas darker. This has similarities to adding more detail - if you have a high amount of detail in an area you might expect adjacent pixels to vary a lot in colour, and if there wasn't much detail you might expect there not to be much difference. But sharpening adds microcontrast, not additional detail. No matter how much you sharpen the roof in our up-scaled 24MP image, you're not going to start seeing tile details that weren't there in the first place.

(snip)

 

Sometimes you can. With non-linear deconvolution, and enough signal/noise, and if you know the spot function well enough, you can do some surprising things.

 

Some of the best examples were done with the early Hubble telescope images, with the defective mirror.

 

The shape of the mirror was known extremely accurately, and so the spot function.

 

They took pictures of brighter objects, to get the needed high S/N, and got some amazing pictures.

I think I'd still argue the Hubble case as a "microcontrast enhancement" rather than changing resolution, Glen - albeit a special case. It was similar to trying to compensate for an out of focus lens (which I believe is what Photoshop's smart sharpen tool does with "lens blur": deconvolution with a flat disk). The Hubble system had the advantage of being very precisely machined, just to slightly the wrong specification - and it helps that its subjects are always effectively at infinity.

Micro contrast is a term that crept into use with its co-conspirator 'acutance' during the 1960s.

 

Well, there is some truth that 'enthusiast' usages of these terms are largely metaphorical or analogical.

 

However, both contrast and acutance are real words describing actual phenomena. A good textbook on optics will be a useful, but difficult, introduction. Wikipedia can point the way too:

contrast

acutance

For what it's worth, I found the explanation of micro contrast very interesting, because up until reading this thread I had no idea what it means.

 

On the never ending discussion of bokeh, I reckon it's blur, but people usually refer to blur as bokeh only if they like the look of the blur. So only a pleasant looking blur is bokeh. Of course that is highly subjective, which is why people can talk about it endlessly.