Practicing the craft: Monochromatic red and white striped lily

[JBW]

That is a rather interesting and unique view Brian.

All of this was considerably less complex and simpler to get great images than colour has ever been.

There are far more variables in colour work, from white balance to colour grading. I find doing colour well takes a lot more time, effort and knowledge (colour theory) than B&W does.

To get great images in colour just set your camera to full auto, or any number of settings. While I have not your vast experience in shooting I have been looking at pictures probably as long as you have. In my experience the most powerful shots have been the ones sans colour. I do believe we shall have to agree to disagree on this one.

[Manfred M]

The best reason I can think of to make a monochrome image rather than a color image is that I like it better. I can't perceive 16 million colors, so it doesn't do me any good that so many of them are being displayed.

You are quite right. The human visual system can only perceive an estimated 10 million distinct colours. Even the very best computer screen can only show a subset of that range.

[Mike Buckley]

The human visual system can only perceive an estimated 10 million distinct colours.

I can assure you that I can't perceive any where near 10 million distinct colors. I can't even count that high without making mistakes.

[Manfred M]

I can assure you that I can't perceive any where near 10 million distinct colors. I can't even count that high without making mistakes.

It's a good thing that your visual system runs on automatic then.

The human visual system runs much like the Bayer sensor in our cameras, with our colour vision coming from a sensor type that is referred to as a "cone". Each eye has about 6 million of these cones with roughly 50% of these sensitive to green wavelengths and 25% to each of the red and blue wavelengths of light.

While 10 million may sound like a large number, what this breaks down to is that the human visual system can differentiate between 215 shades of red, 215 shades of green and 215 shades of blue (not that overwhelming a number, I think). 215 x 215 x 215 = 9.94 million (which is a bit overwhelming).

[Mike Buckley]

the human visual system can differentiate between 215 shades of red, 215 shades of green and 215 shades of blue (not that overwhelming a number, I think). 215 x 215 x 215 = 9.94 million (which is a bit overwhelming).

Try taking 215 shades of any color that are arranged so that no two shades of consecutive color are placed next to each other. Then try rearranging all 215 shades so they are in perfect order from darkest to lightest. Very, very few people will be able to do it. That's because they won't be able to distinguish between all 215 colors, much less 10 million colors.

We know that professional baseball batters can see a baseball thrown by a professional baseball pitcher at speeds up to 100 miles per hour and more. We also know that only 10% of the human population has eyes that are capable of seeing that.

[Manfred M]

To get great images in colour just set your camera to full auto, or any number of settings. While I have not your vast experience in shooting I have been looking at pictures probably as long as you have. In my experience the most powerful shots have been the ones sans colour. I do believe we shall have to agree to disagree on this one.

You can also set your camera to take B&W (jpeg) images. That approach was recommended to the class by my photographic composition professor.

Brian - I'm not disputing that B&W makes for beautiful images; in fact four of my favourite photographers; Edward Weston, Ansel Adams, Yousef Karsh and Henri Cartier-Bresson are best known for their B&W work. The latter three did try shooting in colour (I am not sure about Weston), but they returned to their comfort zone in B&W.

All I have suggested is that in my experience, it is far easier to get a great image in B&W than colour, simply because there are less variables that have to be managed. By removing colour from the process we have applied one of the most important compositional techniques; simplification. Because of that challenge, I tend to work more in colour than in B&W, but I also do go to B&W when I find it is the best approach from a compositional standpoint...

[xpatUSA]

Try taking 215 shades of any color that are arranged so that no two shades of consecutive color are placed next to each other. Then try rearranging all 215 shades so they are in perfect order from darkest to lightest. Very, very few people will be able to do it. That's because they won't be able to distinguish between all 215 colors, much less 10 million colors.

One of the finest tools for playing with colors is Bruce Lindbloom's color calculator which can convert anything to anything:

http://www.brucelindbloom.com/index....alculator.html

Then there is the concept of the "just noticeable difference":

https://en.wikipedia.org/wiki/Just-n...ble_difference (JND)

For color, the units of JND are those of the CIELAB color model; for humans the average JND is often said to be ~2.3 delta-E. Now it is possible to, for example, change the lightness of a color in LAB using the calculator and see how the sRGB numbers change.

I did that, starting with an "equal energy" gray i.e. XYZ = 0.31,0.31,0.31. I noted the RGB values, then reduced the L* value by 2.3 - a difference "just noticeable" to the average human under average conditions etc., blah-di-blah.

The point of this post is that the calculated sRGB values (0-255) changed by a whopping -6, telling us that our monitors are capable of displaying luminance differences that are well below the ability of our eyes to perceive them!

I also went backwards and changed the sRGB numbers by one. The L* value changed by only 0.4 delta-E which is much less than even the best of us can distinguish.

OT, but that makes me wonder why some folks buy 10-bit monitors and why others just can't wait for the Holy Grail of 16 bits and Rec. 2020.

[Wavelength]

Excellent image

[Manfred M]

I can assure you that I can't perceive any where near 10 million distinct colors. I can't even count that high without making mistakes.

Mike - I will leave the answer of that to the human physiology experts.

Age, gender, heredity and disease all impact the human visual system. Our colour sensitivity gets worse as we age, women tend to have better colour vision than men (in the "old" days, colour labs tended to use women as printers for this reason), colour acuity as well as colour blindness are often inherited conditions and certain medical conditions can also change colour vision. Males in our demographic (and let me include Ted in the mix) are unlikely to have peak colour vision.

The issue in defining what normal human vision is about doesn't seem to be getting a lot of academic research. When I see many of the current standards (various CIE colour spaces) that were agreed to in 1931, based on work done by a couple of researchers in the late 1920s. This suggests to me that the sample to determine "normal" human vision was likely based on European subjects and likely a bias towards younger males. I would like to see this aspect of colour vision revisited using more a representative world population model.

[Mike Buckley]

I will leave the answer of that to the human physiology experts.

The nutrition experts told us when I was a child to eat lots of eggs. Then when I became an adult, they changed their minds and told us to dramatically limit our eating of eggs. Now that I'm closer to the end of my life than the beginning, they've changed their minds yet again and tell us that it's fine to eat eggs daily.

That's just one example of why the so-called experts backed by the findings of their so-called science rarely impress me. That's especially true when a statement that the typical human should be able to distinguish 10 million colors doesn't seem to me to pass the test of common sense or my life experiences.

[Manfred M]

That's just one example of why the so-called experts backed by the findings of their so-called science rarely impress me. That's especially true when a statement that the typical human should be able to distinguish 10 million colors doesn't seem to me to pass the test of common sense or my life experiences.

Mike - there is a lot of science that does not pass the "common sense" test. This is quite true when it comes to sub-atomic physics (quantum mechanics). For instance, how can something act like a particle yet have wave-like characteristics? Intuitively when we throw a ball, we have a particle in motion. When we look at waves in a pool, they ripple and we see waves with crests and peaks. Both of those are intuitive and well known principles of classical (Newtonian) physics.

Fortunately for us photographers this is exactly how light behaves (duality of light; actually electromagnetic radiation as it exceeds the wavelengths of visible light) and we can count the individual photons that hit a detector (particle properties) as well as capture the wavelength (colour) of the photons. That definitely does not pass the "common sense test", yet it is still reality.

Science is not static and part of the beauty of science is that our knowledge of it changes as we gain knowledge, we update the models and science moves on. For those of us with a scientific background, this is a key strength of science, not a weakness.

Science can get things really wrong too. For hundreds of years, scientists thought that outer space was filled with a mysterious substance called "aether", that allowed light waves to travel through it. Aether passed the "common sense test" as light from the sun was known to travel to earth. Along came people like Einstein, James Clerk Maxwell and Louis de Broglie and their work (along with others) proved that there was no thing as aether and light was able to travel through a vacuum. Theories (i.e. models were updated) and science moved on.

Fortunately, science seems to get things right more often than it gets things wrong. Our standard of living is much better than what our ancestors had and we lead longer and healthier lives.

[Mike Buckley]

I don't have the technical or scientific background to understand much of what you wrote in your previous post, Manfred. Even so, I appreciate the enjoyable discussion.

I do agree with you that science and/or experts get things right more often than not. However, when they get it wrong, it seems like it's really, really wrong. So, as a non-technical kind of guy, I maintain a healthy sense of skepticism, so much so that I can't bring myself to irrefutably attribute our longer and healthier lives solely to or even mostly to advancements in science. That's because at the very least there are other contributing factors.

Back to my main point: You often write that the reduction of 16 million shades in color to 256 shades in black-and-white is a good reason to not "throw away" all those colors. That just doesn't make any sense to me for the reasons I hopefully explained in this discussion. If it doesn't make any sense, that's fine also!

[Manfred M]

Back to my main point: You often write that the reduction of 16 million shades in color to 256 shades in black-and-white is a good reason to not "throw away" all those colors. That just doesn't make any sense to me for the reasons I hopefully explained in this discussion. If it doesn't make any sense, that's fine also!

Mike I write it because I feel it is an important consideration to be made when deciding to go colour or monochrome.

You lose most of the data that your camera has recorded, and it should be something that the photographer should consider.

That being said, it can be the right decision to get the look one is after. I often write that simplification is a very important compositional tool, and throwing away 99.998% of the data your camera can captured simplifies things a lot!

[Mike Buckley]

[Changing to black-and-white loses] most of the data that your camera has recorded, and it should be something that the photographer should consider.

I completely disagree, as the loss of data in and of itself is entirely unimportant. As an example, we can change from color to black-and-white by converting and using a color filter or we can make the change by desaturating. As you are aware, the latter method results in one-third the amount of data than the former method. Even so, when there are absolutely no disadvantages associated with that loss of data, which can indeed be the case, there is no compelling reason not to desaturate.

[Manfred M]

I completely disagree, as the loss of data in and of itself is entirely unimportant. As an example, we can change from color to black-and-white by converting and using a color filter or we can make the change by desaturating. As you are aware, the latter method results in one-third the amount of data than the former method. Even so, when there are absolutely no disadvantages associated with that loss of data, which can indeed be the case, there is no compelling reason not to desaturate.

Let's agree to disagree on this one.

[xpatUSA]

I completely disagree, as the loss of data in and of itself is entirely unimportant.

Hmmm . . .

As an example, we can change from color to black-and-white by converting and using a color filter or we can make the change by desaturating. As you are aware, the latter method results in one-third the amount of data than the former method.

Not on my computer, sorry Mike. I just now converted a raw file to an 8 bit per channel TIFF. Output file size was 13.3MB. Then I de-saturated it and saved again. New output file size was 13.2MB.

There must be more to the bolded statement than meets the eye.

In other words, the act of desaturation per se does not "lose data" although it does of course change the data.

Perhaps the word "data" is being used a bit loosely in this thread?

[Manfred M]

Ted - You are of course correct from a data standpoint. Desaturating an image will set the R, G and B values to be the same, so the image sizes in TIFF would be the same, apart from any changes in the header data.

Yes you are also correct, the data would be changed, and in that one is losing the original values assigned to each subpixel in the image (unless of course the part of the image was already black, white or some gray value). In the 8-bit example, the three subpixels could represent up to 16.8 million discrete values. The moment we assign the same value to each subpixel, only 256 discrete values are available.

[Mike Buckley]

Ted,

My statement about losing two-thirds of the data comes directly from pages 9 and 10 of Vincent Versace's book, From Oz to Kansas. He explains that "Photoshop keeps all the color/luminance information in Lab and then translates it as necessary to ProPhoto RGB, CMYK, sRGB, Grayscale, etc. To see this, convert an RGB image to Lab and watch what happens in the Info panel. Applying the destaurate command discards the A (green-magenta) and B (blue-yellow) color channels, retaining only the Luminance (L) channel information. Two of three channels are lost, therefore, you have lost two-thirds of the original file's information, and the file size shrinks accordingly. Translating the file back to RGB returns it to the original file size, but what remains in the R,G, and B channels is only the L channel information. The same thing happens if you convert RGB to Grayscale; the file size is reduced by two-thirds."

I didn't check him out on that; I took his word for it. At the time his book was published, all references in it to Photoshop were to version CS6.

[Manfred M]

I suspect that Versace is partially correct, but only so far as Adobe Camera Raw (ACR) is part of Photoshop.

Raw data has no colour space assigned to it, so the most generic approach will be to use the LAB colourspace in the raw convertor. If one looks at the Basic panel in ACR (or Lightroom, for that matter) we see three controls:

1. Exposure - this adjusts the "lightness" of the data only, i.e. the "L" channel.

2. Tint - this is the "A" channel as the green / magenta relationship is adjusted.

3. Temperature - This is where we set the white balance and in sunlight, that means the blue / yellow settings which corresponds to the "B" channel in the LAB colour space.




Once the data is converted into a Photoshop document, as part of the raw conversion process, a colour space is "baked" into the file. After this occurs, it really makes no sense to continue to work in the LAB colour space as there would have to be a conversion every time an edit occurs. The actual calculations for the various colour models are defined by external bodies, so to ensure that the colour models are applied correctly, this needs to be done natively, rather than by working in the LAB colour space. If the "Photoshop uses LAB internally" were true, we wouldn't see the issues of issues colour space conversions, like the well known ProPhoto to CYMK issue.


The other issue is that the internal manipulation of data has nothing to do with the way that data is written to files, like TIFF, JPEG, etc. These files have their own internal structure and all RGB formats store each pixel as three digits, one for R values, one for G values and one for B values, even if all three values are the same. This is why Ted sees little change in file size between the colour 8-bit version and the gray scale 8-bit version. If we were working with a CMYK file, the data would be stored with 4 digits of information for every pixel.

[Mike Buckley]

Manfred,

I'm not familiar enough with Photoshop or ACR to digest the stuff you explained. However, it might be helpful to note that I think Versace was referring only to Photoshop, not ACR. He clearly states that his workflow was to change to monochrome in Photoshop rather than ACR and that he described ACR as a plug-in to Photoshop. He does describe in other parts of the book using ACR, so considering that he didn't mention ACR in the exercise explained in my previous post, that makes me confident that he was referring only to Photoshop.

Having said that, I have absolutely no idea if that information is helpful to you.