Are Bayer Pattern Color Photo Conversions to Black and White Really Unsharp? Really?

 

Store Window in San Angelo, Texas. 

When the world seems to be going to hell in a tote bag B. and I tend to retreat into nostalgia. It's better on the psyche than watching falling markets ratchet down minute by minute and it takes our minds off the reality that straightening out last week's destruction of the U.S. (and world) economy is going to take some time. Maybe a lot of time. We're choosing to spend that time catching up on projects we meant to do anyway. 

One project that we're both focused on is going through my library of thousands of color food images to find some good stuff to redecorate our dining room and our kitchen with. Graphic close-ups of fresh produce. Photos of people shopping the markets. And still life images of stacks of interesting dish ware and utensils. To facilitate the process I converted raw files into Jpegs and made easily accessible galleries on Smugmug of markets in Austin, Vancouver, Montreal and San Antonio. We were surprised that shooting regularly over a couple of decades can add up to a prodigious number of good candidates for room decoration. For everyday usable art. 

But as I was digging into various folders and Lightroom folders I came across a neglected vacation folder. It was a trip from Austin to San Angelo to Roswell, NM, to Santa Fe and back again. It was an odd time. We weren't flying then because of Covid, and our responsibilities to aging parents, so we took a road trip instead. Going by car slowed down the process and that was fine with B. and me. We'd gotten into the modern vacation paradigm of flying off to some place trendy, staying a week and then flying back home. It's a totally different sensibility when one drives. You see the mileage. You register the wide open spaces and you connect, I think, much better with your travel. Just seeing the difference in gas station designs can be a thin. Diners in some small towns you'll venture through are like museum pieces. And it can be so interesting to see what's just around the next corner. 

When I found the folder for the 2021 vacation I decided to take a brief look and see if I still liked any of the photographs. There were no food images taken on the trip so I was taking a detour from the task at hand. It was surprising how not looking at a group of photographs for four years could allow you to see what you shot with fresh eyes. With a different perspective. Looking at vacation photos in the days and weeks after your return home prejudices the assessment. You're still busy thinking how you could have done things differently with your camera. Which lens you should have used. And there is also the urge to process selected images in a way that may not seem right when you look at the images years later. 

There are a lot of color images in the 2021 vacation folder but there are some in there that I played with at the time by converting them from color raw files to monochrome (or, black and white) images. And I liked the results I was getting from simple conversions in PhotoShop and Lightroom. 

But every so often I'll read something on the web that describes dedicated monochrome camera images as much sharper and much more detailed than files which start life as color images in cameras with Bayer Pattern filters. I don't know if my eyes are just faulty or someone else's perspective is colored by their gear investments and their need to be contrarian to the usual processes but I find myself a bit baffled. 

Now, I may not be the sharpest "lens" in the camera bag of photography but I have had an extensive, forty plus year history of printing black and white negatives from every format from Minox 16mm to 8x10 Verichome Pan, and everything in between. But the more important point, I guess, is that I've printed thousands and thousands of prints, in various sizes, over those same years. I have also spent time, as a faculty member at UT's College of Fine Arts taking students to see world famous prints at the Humanities Research Center on campus. Even though I am, of course, prejudiced in my beliefs in my ability to assess the quality of photographic images, I do believe I've spent a lot more time, hands on, than most people. And I wanted to bring that perspective to bear as I tried to understand the difference between what I was reading and what I was actually seeing. How sharpness differences manifested in the "real world." 

One point that I think is vital to understand is the role that sensor resolution plays when making comparisons. While the Bayer Filter sitting across the sensor on full color cameras soaks up the equivalent of one stop of overall illumination to the sensor compared to an unfiltered sensor the whole picture is much more complex. See this engaging article to learn more: https://photo.stackexchange.com/questions/87528/how-much-light-and-resolution-is-lost-to-color-filter-arrays

A thorough reading of several sources shows that while the Bayer filter can cut light to the sensor by 2X or more the actual efficiency of a sensor with a Bayer filter in terms of overall resolution is about 50% or one stop difference. And the visual effect, or what your eye sees, is usually much less because of ever more precise algorithms which use information from surrounding clusters of pixels to interpolate color. 

If you match up a color camera and a monochrome camera of the same resolutions the measured resolution of the color camera is, mathematically, 50% of the B&W only camera. But many photographers are using monochrome cameras with only 24 megapixels of resolution while many popular color cameras in use right now have double the megapixel resolution or more. Even if we use the 50% loss of information as a standard it would mean that a 47 megapixel color sensor would effectively equal the 24 megapixel B&W sensor but would suffer a penalty of a stop to two stops less illumination efficiency.  But importantly the resolution would be the same. While to the perception of the human eye the filtered sensor, coupled with advanced algorithms, could even exceed the perceptual maximum of the unfiltered sensor. 

When I have compared 24 megapixel B&W file apparent sharpness; encompassing detail, accutance and fine texture, with downsampled conversions from 47 megapixel color sensors there are differences in tonality from the sensitivity to different colors in a scene and the panchromatic rendering versus linear rendering the overall sharpness and resolution is, for all intents and purposes equivalent. Obviously comparing 24 versus 24 megapixel sensors gives the advantage to the monochrome sensor but with the loss from the monochrome sensor of advanced color channel control. A loss which yields a diminishing of much of the creative control in the final look of a file. 

While most things can be measured the measurements are generally presented in a vacuum while photographic equipment operates as a system. Much depends on the quality of a lens, the stability of the whole system platform, the method of post processing and even the efficiency of various cameras' de-Bayering algorithms and processing. 

Raw, unprocessed files out of almost any kind of camera from either side of the aisle, are linear tonal constructs and it's only after a characteristic curve is applied to a file that we can see a clear image that appears tonally correct to our eyes. An "S" curve of some sort is nearly always applied in the first step of post processing by whatever program is used. You generally never see an actual, linear raw file on your screen because when you open a raw file in any modern processing application a general curve or a precise curve has been applied to the file upon opening. Since the contrast of an image directly affects our perception of sharpness it is possible that the automatic application of a curve profile is in fact more important to our perception of the qualities of a monochrome image than what kind of sensor was used in the image's initial creation. 

A further step for some photographers is to use the built-in camera profiles for monochrome along with the utilization of Jpegs as a file type. In older generation digital color cameras manufacturers generally took the short cut version of just deducting the saturation of a color file in order to make a representation of black and white. A deeper study of newer generations of color cameras shows that camera makers have created B&W profiles that are color spectrum sensitive and use precise applications of color channels, in conjunction with appropriate contrast curves to more precisely and convincingly create monochrome files that are much more complex and pleasingly close to classic black and white film imaging and paper printing. These profiles take into consideration contrast levels in multiple sectors or tonal ranges of the files instead of depending on just one overarching and averaging contrast curve.

If one were to eliminate the variables that come from the lens, the skill of the photographer in providing a stable platform for the system, and in doing legitimate post processing with the best tools I submit that there would be little, if any, difference in any 47+ resolution color camera film converted to monochrome and an identical but half resolution dedicated black and white camera file. Even though the owners of those monochrome cameras would wish otherwise.

But is sharpness and detail that big of a differentiator in current photography? I think not. Those properties can be cold and analytical in a field where aesthetics is more functionally and fundamentally important than small percentage increases in already overly endowed resolution characteristics. Finally, we have to understand the limitations of viewing distance on perceived sharpness and detail. The human eye can only resolve about 5 to 15 megapixels of information when stationary (eyes not scanning). At a glance. For instance while taking in a whole photographic print. The eyes have an imputed resolution of over 500 megapixels but, importantly, only while moving and scanning. And the eye has no actual pixels and perceives color and detail in a much different way than a sensor. 

When we view a photographic print we are limited by factors such as how bright the viewing lights on the print are as well as our age, our overall health, our personal visual potential and, equally important, the viewing distance to the print and the size of the print. Also, prints have properties of granularity, halation and are often degraded by back reflection and artifacts of internal reflection. The substrate itself introduces limitations in actual resolution as well. Given the complex mix of limiting factors just about any current digital camera resolving 24 megapixels or more, black and white only or in full color, is enough to provide the correct amount of detail given standard and accepted viewing distances. Certainly you can press your nose against a print but that's hardly the common use case for viewing photographic prints. The bigger the print the lower per inch resolution needed to deliver the same viewing perception as a smaller print. 

Many people who are using monochrome only cameras and displaying the results on websites and various screen-centric presentations are making photographs of landscapes, urban constructions, etc. A good proportion of the subjects photographed are highly suitable for yet another "curative" for the interference (pun intended) of Bayer filters. That would be the use of widely available in-camera, multi-shot resolution enhancement. In those (usually tripod mounted) cameras the sensor is moved up to eight times in small increments during the capture time and the eight resulting images are blended together while throwing out anomalies occurring between "layers." An eight step multi-res process adds up to four times the resolution of a single exposure which should yield two times as much unadulterated detail as a regular file. A conversion and downsampling of these files to match the size of a single shot mono file should be more than equivalent in terms of the accuracy of the raw data. At that point the post processing application is the common denominator/inflection point of differentiation. 

As I said at the beginning, I'm not the smartest guy in the gallery but I do trust my eyes to detect differences in final files and prints. That's why I often don't take other people's anecdotal relaying of information as gospel. I prefer to see what the actual effects of mixtures of complex interrelationships of imaging parameters are in the real world. As in: How do the differently generated images look on the screen after I've post processed them? How do the differently generated images look on paper after I have printed them? 

You may see all kinds of differences in different cameras. Those differences may be down to any number of things beyond the kind of sensor being used. While color cameras can make very, very convincing and pleasing black and white images you will find that there are not any one shot, black and white only cameras that can make a color image worth a damn. 

One observational note about the desire-ability of specialty black and white cameras is this: I see more monochrome model Leica rangefinder cameras on the used camera market in a season (as a percentage of total sales) than I do the stock, color versions of the same model family. Used, tested, tried and released back into the wild. Not an exact study but a trend I've certainly seen recur over time. 

Of course, I could be wrong but as long as I am able to pull sharp, detailed, contrast-appropriate images out of Bayer Filtered camera sensors I think I'm okay with it. Everyone is looking for a magic bullet in this artistic field. The Placebo Effect is strong among photographers. How else could Leica and Hasselblad command such high prices??? 

And yeah. I fall for marketing as hard or harder than everyone else. 

I like this image because of the soft gradations of the white sheets in the upper highlight range. The black band on the hat provides a nice tonal contrast as do the details of the weave in the hat. Would this image be better with greater perception of detail or would it be trading off the initial intention of the photographer?

It seems a bit disingenuous to show an image that's 3200 pixel wide when it started life at 8800+ pixels wide and then try to make an argument that it is sharper or less sharp than it would be if made with a different camera. On the original file one can see details on individual bricks. Those are not visible on this presentation. The final use of the file is the important point. Resolution, or post production sharpness used as bragging rights has nothing to do with making images for general viewing. Just as using Kodak's ISO Technical Pan 25 black and white film wasn't and isn't the appropriate substitute for Tri-X film. Not from an artistic/aestetic point of view..... anyway.