WARNING: THE FOLLOWING IS A LONG, RAMBLING POST! IF YOU DON'T LIKE LONG POSTS, SCROLL DOWN NOW!
I have been doing more thinking about Sean's contention that lower-contrast lenses might be desirable on a digital camera. And since it's kind of slow at work between Christmas and the New Year, I had some time today to do some reading and experimenting.
As a result, the whole concept is starting to make sense to me.
TWO KINDS OF CONTRAST LOSS
My initial problem with the idea expressed in Sean's Luminous Landscape review (see exchange above) came from the fact that he seemed to be using the term "contrast" in a different way than I was accustomed to using with lenses (and the way it's used in the
Reichmann article, for example.)
And it turns out that this was the case! From my reading this morning, I've learned that a lens reduces contrast in
two different ways: scattering and
veiling glare. And these two contrast-reducing mechanisms have different effects on the image.
(This doesn't just apply to camera lenses: a lot of the best info comes from the X-ray and photonics fields.)
Here are a couple of definitions I copied from the online
Photonics Dictionary:
Veiling glare - Diffuse stray light at the image plane of an optical system that results in reduced contrast and resolution.
Scattering - Change of the spatial distribution of a beam of radiation when it interacts with a surface or a heterogeneous medium, in which process there is no change of wavelength of the radiation.
The key difference for us is that veiling glare is
diffuse (non-image-forming) while scattering retains its beam characteristics. To oversimplify, you could say that veiling glare reduces contrast by spraying image rays
everywhere, while scattering reduces image contrast by sending image rays to
the wrong destination.
GRAY BLACKS? OR FUZZY EDGES?
You can think about the effect of this difference on photography by imagining a highly magnified view of a sharp black-to-white edge, the kind of "knife edge" detail used in making MTF tests.
Diffuse veiling glare would fill in the black area of the image, turning it to light gray --
but this wouldn't affect the abruptness of the black-to-light boundary.
Scattering, on the other hand, would let rays that should hit the white part of the image stray over into the black part. They'd do this statistically: most would miss only by a little, some would miss by more, and a few would miss by a lot. The effect would be to turn the knife-edge boundary into a gradated "gray ramp."
But, this wouldn't affect the overall dark-to-light contrast of the image, because most of the black area would still be pure black.
At the bottom of this rambling missive, I've attached some example images I diddled up with Photoshop (using a white fill to simulate veiling glare, and Gaussian blur to simulate scattering.)
WHEN GRAY IS GOOD
Sean makes a case in a later reply (above) for the benefits of scattering, but I'm still not sure I'm sold on that one -- I like images with crisply-defined small details such as skin and fabric textures.
But the idea that a certain amount of veiling glare could be helpful is starting to get persuasive for me. The key feature here is that once scattering has reduced contrast by wiping out sharp detail boundaries, there's no way to put them back. But if a certain amount of veiling glare has reduced contrast by lightening dark tones, it's pretty easy to darken them again.
This raises the possibility, as Sean suggested in his review, of choosing lenses that reduce contrast enough to help the scene's brightness range fit the limited capabilities of a digital camera's sensor -- then "reconstituting" the lost contrast later by post-processing with Photoshop or another image editor.
But if this is such a great idea, you might ask, why do lens designers strive to reduce veiling glare in the first place? And why would they tolerate scattering at all, if it obliterates fine details? And (most interesting to us as RF users) why is it that some older lens designs seem to be especially "digital friendly" when we stick them onto an R-D1?
Well, I've got a theory about all this, based on the history of photo technology.
HARD CHOICES IN LENS DESIGN
First, let's concede that all lens design is a matter of compromises, and that someone sitting down to design a new lens accepts that some contrast loss is unavoidable.
Let's also assume that a savvy lens designer (nowadays with the help of a computer, in the old days with the help of experience) can choose design options that would tolerate
more contrast loss via veiling glare to get
less contrast loss via scattering... or vice-versa.
Now, here's the guts of my theory. At the beginning of the "RF era," most important photographs were made on black-and-white film. In black-and-white, it's easy to control final print contrast through film development or printing controls.
So, designers of the classic old RF lenses would be willing to tolerate some contrast loss through veiling glare -- that could always be fixed later. But they'd try to avoid contrast loss through scattering, because that could
not be fixed and would wipe out the subtle textures and microcontrasts that make black-and-white photos beautiful.
The ultimate result of this kind of thinking would be a lens such as the old-design Summicron that Sean mentions in his review: one with high detail sharpness but only moderate image contrast.
Okay -- if low contrast/high sharpness was such a great approach, why would designers ever change it?
One-word answer:
color. The first popular color films were slide films, and slide film gives the photographer
no control over the final image's contrast: if the lens doesn't get it onto the film, it ain't gonna be there. This didn't change much with the popularization of color-negative films, since they don't respond to the same kinds of print-contrast controls we can use in black-and-white.
Now, it's only a theory, but I submit that this is the factor that led design houses to move in the opposite direction: accepting more scatter to get less veiling glare, pushing more and more for lenses that produced images with high
dark-to-light contrast.
It made sense: photographers wanted "snappy-looking" color slides and prints, and since color film is inherently less sharp than black-and-white anyway (more emulsion layers) there was no point in sacrificing snap to retain tiny textural contrasts.
THE CLOCK TURNS BACK
But now, digital imaging has turned back the clock! Once again, it's easy for the photographer to control the contrast and tonal distribution of his final image -- just as it was in the days when any serious photographer developed his own film and made his own prints.
With that shift, the unique optical characteristics of classic RF-era lenses have become relevant again.
And for now, the
only way for a digital shooter to exploit those unique characteristics is to use an R-D1! (Neener, neener, EOS-boy...)
LET'S TALK CONTRAST
What we need now to take advantage of this position is to agree on a way to describe some of these characteristics. For example, when you say you've got a "low-contrast" lens, can you distinguish whether you mean it has low
detail contrast (via scattering) or low
dark-to-light contrast (via veiling glare)?
Think about it. Maybe someone can come up with a few ideas, and we can start compiling lists of what classic and modern RF lenses have especially digital-friendly imaging traits...
Blacker, or edgier? Pick your poison...
Here are a few notes on the attached sample image. I used a "fake subject" in the form of a stepped gray scale generated in Photoshop; this is the middle band of the image.
The two bands above it show reduced
detail (or edge) contrast via scattering, which I simulated using the Gaussian Blur filter.
The key thing to notice is that this doesn't change the
tonal (dark-to-light) contrast -- the blacks are just as black as on the original center-band image. What gets lost are the edges between the steps.
The bands below the middle one illustrate tonal contrast reduction via veiling glare, which I simulated with a 20% white fill. You can see that the tonal contrast deteriorates (fogged black steps; think how terrible this would look in a color slide!) but the edge contrast is just as high as the original.
And, as shown in the bottom band, the original tonal contrast can be reconstituted by adjusting the Levels control (just as the wet-darkroom photographer would do by extending the film development time a touch, or printing with a higher-contrast filter.) The statistics of this image are just about identical to those of the original, showing that no actual information was lost.
Now, on to my next project: Hanging all my strange old LTM lenses on my R-D1 and finding which ones have the desirable (for me) combination of low scattering and moderate veiling glare...
