Digital Seeing Hearing Understanding.
How Much Bit Depth For Monochrome?
by Uwe Heimburger
Overview
Since a long long time I'm loving black & white photography.
The german Wikipedia article about "Grau" [4] (engl. Grey or Gray) tells us, "Ein Prozent der Deutschen wählt Grau als Lieblingsfarbe. Es sind fast ausschließlich Männer, die von Beruf Steinmetz oder Informatiker sind." ("engl. One percent of Germans choose grey as their favorite color. These are nearly only men with profession mason or computer scientist.") - I studied computer science (Informatik) ...
Finally, I got a digital camera that uses a monochrome sensor. Monochrome means, this camera can only capture brightness as tones of a graytone scale, often referred to as black & white. A toning option is availlable, so the camera is able to produce JPG images using a specified color for toning.
From now on I'm consistently using the term monochrome instead of black & white, since the basic concept is about using a graytone scale and not only the two values black and white. Of how many values this grayscale consists, depends on the bit depth of the sensor and the image format we produce from it.
In general the digital monochrome grayscale consists of a lot less values than a color sensor will offer. So compared to the black & white film equivalent, which only is restricted by molecular structure and density, or digital RGB color, variety is very limited.
The interpretation of monochromee photographs will be mostly different to the interpretation of photographs relying strongly on color. Light and shadow contrasts, structures and abstraction are strong drivers for monochrome images. In art monochrome has a strong appearance.
As a monochrome photographer I'm interested in publishing my work in best shape - on screen and as prints. Long ago I asked a fine art printing service in what format I should deliver my monochrome images for printing. The answer was that JPG, i.e. 8 bit data, would be sufficient. To tell the truth, I was not really convinced that this would be true.
So, I searched the internet but didn't find any concrete numbers of how much graytones we human beings are able to distinguish. Such a number would be my guideline to decide about the format I would deliver to my fine art printing service for monochrome prints.
Some days ago a similar question was posted in a PentaxForums.com thread. It was about the kind of monitor would be best for monochrome image development.
I wrote a large post about my thoughts and once more searched the internet for concrete numbers. I found a side note within a Wikipedia article about "Fensterung (Computertomographie)" [1]. This is an important topic in the world of medicine diagnostics (Mammography). So I reported these numbers in my PentaxForums.com post. It didn't take long and another forum member posted that the number he knows of is 700 - 900 graytones - and he could offer a concrete scientific article with internet reference [3].
This numbers sound much more reliable to me than the ones I found in the Wikipedia article. Therefore, from now on I take this numbers as my main guideline.
What Monitor To Use For Monochrome Image Development?
These are my written down thoughts about monitors to use for Monochrome development.
Some mostly technical based thoughts ...
- Today computer software and monitors work based on the RGB Color Model.
- I don't know any computer / photo application that will work based only on a grayscale channel. At the end of the 1980s I worked on an Atari Mega ST. The standard monitor was black or white for the pixels - so only one bit per pixel, the image was dithered. Because of the high densitiy of pixels the eye/brain combo created a usable image of 640x400 Pixels (if I remember right).
- Todays best monitors work at 10 bit RGB, i.e. 10 bit per color component, standard ones at 8 bit.
- For pure monochrome in the RGB model the grayscale is build by R=B=G. So every channel (color component) is the same.
- If we want to be able to use colored grays, i.e. be free in choosing a monochrome tint, we need a color monitor.
- The camera (Pentax K3 Mark 3 Monochrome) delivers 14 bit max. for the one grayscale component, i.e. values from 0 - 16383 (max. value 214-1).
- The monitors theoretical ability is 0 - 1023 (max. 210-1).
- JPG format is only 8 bit, i.e. 0 - 255 (max. 28-1).
- According to a Wikipedia medical article I read [1], our eyes/brain combo can distinguish only 60 - 80 graytones. And from the 8 bit grayscale we are not able to distinguish at least the lower and upper 16 values of the 8-bit scale. So in reality we talk about a max. domain of around 220 values, where we are not able to distinguish direct neigbours. Since the info was only a side note, I don't know the preconditions for this data.
- A PentaxForums.com member reported in his post [2] about 700 - 900 distinguishable gray values. He offered a link to the medical data base article [3], that describes how these values are obtained.
- Apart from that, our individual physical eyes/brain combo will determin what we are able to really distinguish. Main contributers for recognition grays are the rods. The cones that support color recognition are much less sensitive.
This means, if we do some work on our grayscale images like using gradation curve and burning and dodging we should use the 14 bit raw. This will prevent from banding in very smooth gradation areas.
Monitor Usage - Conclusion
So, what are my conclusions about a monitor?
- a 10 bit monitor will offer max. 1024 (grayscale) different values per pixel that fully covers the values our eye will be able to distinguish. 8 bit won't cover the gray values we can distinguish so a 10 bit monitor is the way to go. Apart from that 10 bit gives some room for monitor adjustment / calibration - this seems even more critical to me than with color image development
- an optimal monitor for black & white needs brightness uniformity across the monitor area (remember, gray tones is about brightness values)
- the critical aspect IMO are the blacks. If a monitor cannot show differentiated blacks, which depends on physical properties, work on shadows in the images will be problematic
- monitors that are good at showing differenciated shadows need to be calibrated for optimal grayscale balance since you need to be able to seperate low graytone values as much as your individual eye/brain is able to
- the brightness of the monitor / environment in which we develop our monochrome images might play a bigger role than with color images
For the critical medicine area there should exist most advanced grayscale monitors for diagnostics. But they will be expensive, I guess. So, I think at the end the best solution seems to be a very good usual 10 bit color monitor for monochrome image development.
Printing Monochrome Images - Conclusion
Printing quality depends on the quality of the device we use. Today we get high end fine art printers for an acceptable price and they will offer the quality their big brothers and sisters offer. They are able to fully support the Adobe RGB color space. So they also will offer a lot more than 256 graytones that 8 bit can deliver.
But for print 8 bit might be sufficient. Even after Medex's info about 700-900 graytones we can distinguish.
When we talk about printing we mostly also talk about ppi resolution and viewing distance. With growing viewing distance a pattern of points might merge to a clear image that shows us something meaningfull in sufficient quality. So 8 bit still might hold as a rule for printing monochrome ... on the other hand, I think it can't be wrong to go with 16 bit for monochrome prints, even if there's some waste of megabytes.
I'll choose 16 bit TIFF for printing monochrome images in the future, even if our eye-brain combo can't resolve all the possible fine gradations - highest quality ... if the printing device is able to deliver it. For internet presentation JPG seems to be mostly sufficient to me.
Final Notes
These are my personal conclusions for monochrome image development and printing monochrome images I make using my Pentax K-3 Mark III Monochrome. Of course the same goes for my Color to Monochrome converted images.
Others might follow different standards and may come to other conclusions. Feel free to go your own way! - I hope the offered info does help!
Sources
[1] Fensterung (Computertomographie)
[3] National Library of Medicine, Increasing the Number of Gray Shades in Medical Display Systems—How Much is Enough?, Tom Kimpe and Tom Tuytschaever, Published online 2006 Dec 29: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3043920/