No longer a newbie, moving up!
May 24, 2012
Reaction score
What is Micro Contrast? | DSLRBodies | Thom Hogan
You might already notice that the mountains at the bottom—the Cuernos in Torres del Paine National Park in Chile—don’t seem to have a lot of contrast to them.

Indeed, I’d guess that almost any so-called expert on lens rendering would immediately say that there’s very little micro contrast.


Wait, what happened here? (A Structure tool. Most B&W converters have them.) Look at all that detail in the wall now! Lots of micro contrast, it seems, though the noise in the sky is starting to be a problem (mostly because the red channel has very little value to it, so the more extreme processing is starting to emphasize the noise in that channel). So, wait, is micro contrast a lens thing or a post processing thing?


Then we have the demosaic. For years Nikon DSLR owners have been saying the same thing about Capture NX processing versus Adobe ACR: the Nikon processing has better color and a bit more contrast in a straight conversion. Some of this is the spectral choices, some of it is that Adobe assumes a linear response between two test points, some of it is just that Adobe doesn’t have all the data that Nikon does. A lot of Adobe users almost automatically crank up the Clarity slider. Why? Because it increases the so-called micro contrast in the mid-tones.

Now don’t get me wrong. I’ve long seen that some lenses do indeed produce more useful and easily observable contrast on small tonal differences than others. Those lenses tend to be ones with simple optical formulas, which suggests that both air-to-glass impacts as well as trying to do too much with things like aspherical polishing and how the optical formulas sometimes try to stretch or compress the optical path—as often happens in zooms—must be to blame.

But micro contrast is a much more nuanced problem than that. As I’ve tried to show here, even a lens—and position on that lens—that would be regarded as poor in micro contrast is still passing useful contrast info to the sensor.


You’ll note that in my lens reviews I rarely mention micro contrast. I think you know why now. Trying to distinguish between what was the lens’ contribution and what was the demosaic’s contribution (or deduction) is a fool’s errand. I do sometimes mention micro contrast with a lens when it is so exceptional that it can’t be ignored and must be an attribute of the lens, though. The Zeiss Otus lenses are one good example of that.

What I know is that a modern full frame sensor gives you about 14 EV of dynamic range, but only about 8 bit of color resolution (or, to say it with DxOMark, they provide about 24 Bit color depth; color depth being the total of color resolution over all three color channels). The later is because of the noise of the light itself. That means light is distributes ultimately completely randomly. The random will even itself out if you record more light - but theres always a rest thats noise and kills the precision of your color information.

If your lens has stray light and/or blocks certain parts of the color spectrum more than others, it effectively kills color information, too. That information, in my understanding, is gone and cannot be recovered. I understand that lenses with good micro contrast are those who have very little such stray light.

Thats my current theory about what microcontrast even means.

Most reactions

New Topics