Crazy sharp? MTF, contrast, and sharpness

 

29-9-2014

 

What do we mean by saying "these binos are crazy sharp, extremely crisp and have superb contrast"? How can we analyze how great the optics of a bino really are?

Technically the most solid determination of optical quality relies on MTF measurements. Modulation transfer function (MTF) measurements mean that patterns of black and white bars with known contrast ratio (difference in brightness) and in decreasing sizes ("frequency", often referred to in linepairs per millimeter) are projected as an image by the optical system which is evaluated. The contrast in the image can be measured. The closer the image contrast is to the object contrast (the test chart with black and white bars), the better the optical system is. Surely the finer the bars, the bigger the loss in contrast will be in the image.

In measuring MTF we are looking at how good a lens system transfers contrast. Contrast transfer is a good short term for this key factor in optical performance. So, it´s mainly about contrast.

 

Sharpness - why do we bother

Sharpness is sometimes defined as acutance or edge contrast. A good example would be images of tree branches or the wall of a house against the sky. If we see a good contrast (difference in brightness) along the edges of those branches or this wall in an image, we would say it is a sharp image.

Why are we interested in sharpness? Why do we enjoy sharp images? Our primate ancestors probably depended on quickly judging if a branch would be strong enough to support them, so their visual system evolved to put a lot of emphasis on rather big structures in their habitat, not finest details. On a perceptual level the contrast of not too small edges was important. In fact, our brains increase this edge contrast by the ingenious mechanism of lateral inhibition.

Indeed, if we watch an image from a given distance, our initial perception of sharpness is determined by the edge contrast of coarse things, not the fine ones! Sharpness basically depends on contrast transfer at coarse structures. This was empirically proofed for example by the studies of Otto R. Schade and Erich Heynacher. If we take a binocular and watch a testchart, it will be the bigger bars that will give us the impression: this is sharp. That also explains why a lot of binos, even cheaper ones, can seem very sharp at first glance. For pictures, graphs and more details on our perception of sharpness PLEASE READ page 8 and 9 in this simply superb article from Dr. Hans Kiening from Arri.

If you have looked at the MTF charts manufacturers publish for their photo lenses, you will typically see a contrast transfer of almost 100% for 5 linepairs/mm and a contrast transfer of maybe 60% for 30 linepairs/mm in good lenses. Canon´s Larry Thorpe claimed for their professional broadcast lenses made for 1080p HD cinematography that they wanted to have at least a 50% contrast transfer at a resolution of 50 linepairs/mm. This is hard to achieve and costs money.

Another question is what contrast we need at a given frequency (linepairs/mm) to judge these details to be "sharp". Panavision, the company that supplies Hollywood with cameras and lenses once claimed that it should at least be 30%. They of course were primarily thinking about cinema and HDTV audiences.

Back to binos. If after our first impression we look closer, we might notice that in inferior binos the contrast transfer in our image will rapidly decrease the finer our details get. This will leave us unsatisfied. If on the other hand a bino image even at the resolution limit of our eyes still has a high contrast, it will give us a very rich, vivid, vibrant and satisfying viewing experience. This certainly needs a huge effort in optomechanical manufacturing and makes our binos expensive.

 

Manufacturer´s fingerprints?

To my knowledge it is not possible to artificially raise the contrast transfer of an optical system (given that colour reproduction should be accurate). But it is likely that different manufacturers have different philosophies about which parts of the MTF curve they want to optimize. I´d say that for example Nikon and Zeiss were often going for very high MTFs at low frequencies, yielding proverbial crispness, whereas Leica and Olympus were manufacturing for good contrast at fine details, too. In the times of slide film I was once testing 28mm primes of Nikon, Zeiss and Olympus. All looked reasonably good in projection with the Zeiss and Nikon having a bit more "pop". When I scrutinized the images of the Zeiss and Nikon lenses with a 40x loupe, the details were blurred far above the size of the film grain, whereas in the Olympus images only the film grain itself would determine the resolution of details. The Olympus lens images may have looked a bit less spectacular at first sight, but had a much richer, more detailed image at a closer look.

 

Maybe it´s better not to look too close?

A very similar thing might happen even with top class binoculars, as some recent quarrel about the 54mm Zeiss HTs suggests. Some demanding astronomers were putting samples of those binos on tripods and using 3x boosters to zoom into the images of their refined testcharts. They were not satisfied at all. Beyond sample variation and poor quality control it could simply be that Zeiss optimized these binos for their main customers - hunters and ornithologists who use the glasses 99% handheld and who need a high contrast at medium size objects, not at the size of pinpoint stars. Even if they should use tripods occasionally they surely would not zoom to 1/9th of the image circle and scrutinize it... Sometimes it´s better not to look too close, because the steep price point may fool us into believing we can expect a superb performance under any circumstances.

 

"The Nikon 8x32 SE are the sharpest binos in the world."

Some enthusiasts have claimed this and they may be right - at the big structures which determine our impression of sharpness the SE must have a fantastic contrast transfer. It´s extremely sharp, in fact almost oversharp, because our brain even increases this contrast. But if we look closely at small details, like dewdrops in a meadow, and finally approaching the resolution limit of our eyes, a Habicht 8x30 will probably give a higher contrast transfer. That is one reason why Habicht images look so good, beyond the mere impression of sharpness, and why Nikon SE images seem a little bit sterile and lacking in comparison. Some photographers would call the contrast of fine details the "microcontrast" of an image, quite a poignant term (but probably a false one if you ask a specialist of optical design). One could say that contrast transfer of big things defines shapes, whereas contrast transfer of fine details makes textures vivid. If you should ever test a first class binocular like the Habicht 8x30, make sure to have a look at water - dewdrops, puddles, a river - and appreciate the textural richness... Even a bino with great contrast transfer needs contrasty objects to shine. Don´t test it on a foggy day.