Color management myths and misinformation video

[digitaldog]

For the purpose of this debate (which wasn't meant to be a geeks' orgy, btw), this makes as much sense as saying to someone who asked you how far away is the nearest town: "Sorry, Sir, that's a nonsense question. Everybody knows that distances are conceptually continuous, therefore it's wrong to talk about number of miles or kilometers."

The purpose of what debate? And who says it wasn't meant to be anything but clarification of terminology (does Adobe RGB (1998) have more colors than sRGB).
If someone asks you: How much does your image in sRGB weight in grams, that IS a nonsense question and the statement: Sorry, Sir, that's a nonsense question is perfectly appropriate.
You seem to still have issues either understanding the difference between gamut (range) and number of colors or just want to disagree with what is pretty much a consensus here. You first asked "are there any images that fall outside Adobe RGB (1998)"? or was it sRGB and the color science illustrated the fact that yes, that is possible. You then asked about the difference between color gamut in terms of range and colors, using the two pixel example and that was corrected. You asked about two color numbers in differing locations on a gamut plot and it was explained. I'm not sure what your point of posting here is unless you are still unclear about what color gamut is and what it defines. Are you?

Lastly, when someone writes Everybody knows... The first thing I do is put my BS detector on high. For example: Everyone knows what Gamut means. Everyone knows that gamut and number of colors are two different attributes. The last 10 pages should be proof that both statements are false.

[Slobodan Blagojevic]

... I'm not sure what your point of posting here...

Andrew, would you like me to stop posting here? Stop asking questions?

Let me remind you what you said about the purpose of the potential video:

... to present to someone who is looking for the crux of the facts. It isn't intended for this audience.

Well, Andrew, I am your resident dummy, I am someone looking for "the crux of the facts," explainable in plain English, and in as few sentences as possible.

So, do you want me to stop, Andrew? So that you can continue explain it to each other, until you find an explanation that only satisfies you (as a group)? An explanation that would still leave us, dummies, as perplexed as before? An explanation that would require us, dummies, to read this 10-page (so far) thread, plus the 27-page one, plus watch a 47-min video, plus obtain a PhD in several disciplines, so that, ultimately, we get at the same level as you guys and ultimately understand what you are talking about?

P.S. Actually, Andrew, feel free to ignore my questions. If I do continue to ask them, I hope there will be still some kind souls left to answer them

[digitaldog]

Andrew, would you like me to stop posting here? Stop asking questions?

Not at all! I'd like you to attempt understand and then verify back to us what you've been told rather than argue about what you've been told. Big difference IMHO. I pointed out three posts you made on the topic that contained answers, you didn't even verify you read or understood them. You just argued with some about what you've been told which isn't useful for either party.

[Slobodan Blagojevic]

Not at all! I'd like you to attempt understand and then verify back to us what you've been told rather than argue about what you've been told.

Sounds more like a religious dogma to me. Just listen what we say, repeat, and do not question it. You got to be kidding me, Andrew! That is exactly why I do not go to church.

You do understand that any process of attempting to understand includes further questioning? Are you familiar with the Five-Why technique (i.e., probing initial and any subsequent answers with further "why"s until you get to the bottom of the matter)? Yes, it can drive people crazy, but if you can not answer it, you do not understand it enough yourself.

[digitaldog]

Sounds more like a religious dogma to me. Just listen what we say, repeat, and do not question it. You got to be kidding me, Andrew! That is exactly why I do not go to church.

Just the opposite considering everything discussed here is based on science (color science). The guy you attempted to defend on that closed post was the fellow with the religious dogma. His information about color was technically so off base and grounded in religion not science that it was a bit shocking to see you come to his defence.

You asked at least three questions here about color gamut and got a number of answers. None were challenged by your other peers, only you, the person asking the question have provided push back. I'm not sure why but I'm starting to believe the way your mind works on this topic isn't that far from Gary's.

So I'll ask again, just so you see that I am interested in your further understanding of the topic: I'm not sure what your point of posting here is unless you are still unclear about what color gamut is and what it defines. Are you?

[Slobodan Blagojevic]

... I'm not sure what your point of posting here is unless you are still unclear about what color gamut is and what it defines. Are you?

Yes. And that is why I am still posting here.

[digitaldog]

Yes. And that is why I am still posting here.

Well what is unclear?
Seems that after 10 pages, saying or suggesting one working space has more colors than another is folly. Don't go there.
Do you understand range of colors in terms of the differences in color gamut between these working spaces?

[sandymc]

For the purpose of this debate (which wasn't meant to be a geeks' orgy, btw), this makes as much sense as saying to someone who asked you how far away is the nearest town: "Sorry, Sir, that's a nonsense question. Everybody knows that distances are conceptually continuous, therefore it's wrong to talk about number of miles or kilometers."

The thing is, as regards gamut, there is no "miles or kilometers" or any generally accepted thing that you could count - at least not in any sense that would be relevant to a non-geek. To count things, you have to be counting individual values, which immediately means you're in a particular file format.

Sandy

[Slobodan Blagojevic]

... I  have come to these conclusions...:

1) All the RGB working spaces have exactly the same number of addressable color values and the total number is set by the bit depth of the image file, i.e., 8 bit, 16 bit, etc.).

2) All of these discreetly addressable values are visually discernible values but with three important caveats:  a) To be discernible each addressable color value must be applied to enough adjacent pixels in order to cover an area in the image that subtends a large enough viewing angle to be resolved by the human observer and b) Appropriate surround conditions must be met, i.e., the surrounding near neighbor color values are sufficiently different by at least 1 JND (just noticeable difference) to the human observer, and c) the surrounding color values must also cover image area that can be easily resolved by the viewer.

3) That leaves us with the important concepts of RANGE and SCALE (thank you Andrew).  ProPHoto covers a larger chroma range than aRGB and aRGB covers a larger chroma range than sRGB. The range of hues and range of lightness values remains the same in all of the working RGB color spaces.  Thus, Chroma is the only color property that gets expanded to a larger range and this is done by assigning higher numeric values on the  a* and b* scales when transforming the RGB triplets to CIELAB values.

4) The consequence of encoding a larger chroma range is that precision between the discreetly encoded color values is reduced, hence, the often recommended advice to use higher bit depth, especially with ProPHoto, to ensure that image editing software can manipulate the image tones and colors with sufficient mathematical precision to avoid subtle banding or posterization effects in the final image reproduction.

That's my best shot. Be gentle

best,
Mark
http://www.aardenburg-imaging.com

Hi Mark, this seems like something i can understand. Allow me to see if I got it right, by simplifying it even more, yet remaining still reasonably accurate. What follows is a draft answer to a combined question: "What is the difference between various color spaces, does Adobe RGB have more colors than sRGB, and why my Adobe RGB files appear dull and muted on some displays and in some printouts?"

All RGB spaces have the same number of colors. They differ in their ability to display vivid colors, with wider spaces being able to display a wider range of colors, i.e., more saturated colors. If wider spaces display muted colors, blame the human error, not the color space.

[digitaldog]

Right, good point. ProPhoto does indeed have encoded values that become imaginary on a theoretical basis. I do concede that point. However, those values are forced by any display system for viewing the image inbounds by one method or another. Hence they become discernible, albeit with the caveats I listed before to be met for that imaginary-turned-real color value to be useful and discernible in the image.

This seems to be yet another data point for completely separating number of colors from color gamut.
If I understand most of what has been discussed over these pages in terms of this debate, that being, Adobe RGB has (or hasn't got) more colors than sRGB or X working space has more colors thay Y, it seems quite pointless to try to link the two.

1. As Mark wrote early on, AdobeRGB and sRGB are just color spaces, containers. They don't inherently have any information other than their specifications for primaries, white point, gamma. Until you actually have a pixel, there isn't any information. They seem to have a gamut volume.
2. For something to be a color, we have to be able to see it. How that color is presented to us plays a role too I'm sure. We have at least one well known working space that contains, is a container for numbers representing a 'color' we can't see. That alone would suggest it's rather pointless, perhaps impossible to apply a number of colors onto that color space.
3. Even in a working space or color space that falls within human vision, do all agree what constitutes a visible difference from one color value to the other? Less than 1dE? With what formula?
I'm probably missing other factors as well. But those three are significant enough to convince me the question can and should be answered: it depends on the image and on the encoding. Without either specified, the question is not appropriate and there can't be an answer anyway.

It's a bit like asking, how much does the state of New Mexico weight? We can precisely define and understand the size of the state in square miles or if we wanted a greater precision, square inches. But how could we come up with it's weight? How deep into the earth are will willing to go? Does this count people residing in the state and at what point do we decide how many people are in the state when we weight it? Does this count animals and buildings or just the weight of the stuff that makes up the first 10 inches of land? Or 10 feet? In the end, it is pointless and further, if what we really want to understand is the size of the state of NM, why even discuss another metric, it's weight?

[digitaldog]

"What is the difference between various color spaces, does Adobe RGB have more colors than sRGB, and why my Adobe RGB files appear dull and muted on some displays and in some printouts?"

Adobe RGB doesn't have more colors. And it doesn't appear dull unless you mishandle it, treat it as something it isn't (like sRGB).
If you take sRGB and treat it like Adobe RGB or Adobe RGB and treat it like ProPhoto RGB, it looks too saturated.
You can test this yourself in Photoshop using the Assing Profile Command. Open any color space, even an output color space. Assign the wrong profile. Wrong would be any profile other than the one that correctly defines those RGB values. The numbers do not change when you Assign a different profile. The scale does and that scale is used by Photoshop to provide a correct preview. Adobe RGB (1998) could be treated to look dull or the opposite! Guess what would happen if you assigned a profile that had zero chroma (saturation)?

Here's another analogy for you. You are building a wooden fence that you need to be exactly 30 feet wide. Instead of measuring with a device that provides you feet, you use a device that provides meters. Why is the fence the wrong size? Is the use of wood the issue? No.

[digitaldog]

All RGB spaces have the same number of colors. They differ in their ability to display vivid colors, with wider spaces being able to display a wider range of colors, i.e., more saturated colors. If wider spaces display muted colors, blame the human error, not the color space.

Much better!

[digitaldog]

So based on the comments on color gamut vs. color space, how does this sound? I've obviously done some picking pieces from here and there to come up with this. It's got nothing to do with a video per se, but more of a distilling of the last number of pages on color gamut vs. color numbers:

Does  Adobe RGB (1998) have more colors than sRGB? No. But to uncover why, we have to look at a few facts about color spaces, specifically RGB working spaces like sRGB, Adobe RGB (1998), ProPhoto RGB.

Adobe RGB (1998) and sRGB are just color spaces, containers. They don't inherently have any information other than specifications for primaries, white point, and gamma. Until we actually have a pixel, they don’t contain any information. The pixel has what is called an encoding which can provide a number of possible color values. For example, 24 bit color, (three channels, 8-bit each) can mathematically define 16.7 million color values. Can we see 16.7 million colors? No. Far less. The number is up to debate but the point is, we can use math to produce a value that has no actual relationship to what we can see. All the RGB working spaces have exactly the same number of addressable color values and the total number is set by the bit depth of the image file, i.e., 8-bit, 16-bit.

Before we can define a number of colors, we have to define: what is color? Color, is a perceptual property, something that occurs deep inside our brains. So if you can't see it, it's not a color. As such, colors are defined based on perceptual experiments. Color is not a particular wavelength of light, It is a cognitive perception. We can use math and a metric called deltaE to define when one set of color numbers are imperceptible (indistinguishable) from another set of numbers. For sake or argument, let’s say in one color space, it isn’t possible to see a difference between R0/G78/B0 and R0/G79/B0. As such, we can’t count that example as being two colors, we can’t see any difference between them, they look identical.

Now we have to look at color spaces like ProPhoto RGB. If you examine a plot of this synthetic color space on top of the gamut of human vision, part of it falls outside the plot. It can define numbers which represent colors we can’t see. So these “imagery colors” can’t be counted when we ask, does ProPhoto RGB have more colors than sRGB or another color space. One of the best explanations of why it is folly to even attempt to put a number (of colors) on top of a color space comes from Graeme Gill the creator of the Argyll Color Management System: Colorspaces are conceptually continuous, not discrete, therefore it's wrong to talk about number of colors. Just examining ProPhoto RGB further illustrates it’s impossible to define the number of colors it can contain as it can contain values that we can’t see as colors. Just as 24 bit color can define more values of color numbers than we can see. Encoding is however a useful attribute when editing our images so the point isn’t to dismiss it but rather point out, it provides numbers for something that isn’t a color, it’s just a number. As an analogy, if you were to purchase a ruler to measure something, it is possible the tiny lines that divide up the unit of measure could be finer than you can see. What would be the point of giving you a 1 foot long ruler where there were lines that defined the distance between each was a micron apart instead of a 1/16 of an inch? The micron unit is valid. You can’t see it or use it with your naked eye to measure anything. Think of the encoding of a pixel value the same way with respect to color expect unlike a micron that does exist, a value defining a color you can’t see doesn’t exist; it’s not a color.

The difference in color gamuts is their range and the scale of colors, not the number of colors. This confuses many people because they see a larger gamut plot, a larger volume, and assume larger means more colors. But one has nothing to do with the other. ProPhoto RGB covers a larger range of chroma (what some call Saturation) than Adobe RGB (1998). Adobe RGB (1998) covers a larger range of chroma than sRGB. This has nothing to do with the number of colors, that’s an attribute of how we encode the pixel values. And we can use finer ways to divide up this data. For example, in 16-bit color, the math allows us to define billion’s of color values, but that doesn’t change the fact we still can’t see 16.7 million colors in the 24 bit encoding of these pixels. As such, it’s best to talk about encoding having a potential to define millions or billions of numbers that could be associated to a color if we could see them. But if we can’t differentiae them visibly, it is kind of silly to suggest they are indeed colors. Don’t confuse a color number for a color itself, a color you can see!

[bjanes]

Yes! The green isn't the same. The scale is different even though both use G255 to define this green. 1 yard and 1 meter are not the same distance, the scale is different. The number (1) is the same. G255 in sRGB is a different color than in Adobe RGB (1998). The plot you provided, that Gary can't understand or use, shows this quite clearly. That horseshoe shape represents human vision. One appears more saturated than the other. But they have the same number of colors. Color numbers without a scale (the color space) do not provide enough information to define that color. If you ask me how far I live from you and say 389, am I talking miles, kilometers, lightyears? And none of this has anything to do with the number of colors.

Very well stated. As far as I know, the sRGB specification (and specs for other color spaces) does not specify the encoding format for the values. Integer encoding with a bit depth of 8 or 16 bits is often used, but floating point could also be used. Photoshop allows integer 8 or 16 bits, but also 32 bit floating point, which is usually expressed in the range of 0 to 1.0. The Photoshop info pallet allows values to be expressed as either 8 or 16 integers or in floating point notation with 3 significant digits, and I think this is what the percentage in Lightroom also is using instead reporting as values 0..255 or 0..32769. This actually makes more sense to me.

Indeed, floating point notation in the range of 0..1.0 is used for gamma calculations and other purposes. With floating point notation, the number of encoded colors can be arbitrarily large. In this case, the number of colors contained by the space is best expressed in terms if discretely perceived colors with some sort of delta e consideration.

Bill
 

Bill

[digitaldog]

The Photoshop info pallet allows values to be expressed as either 8 or 16 integers or in floating point notation with 3 significant digits, and I think this is what the percentage in Lightroom also is using instead reporting as values 0..255 or 0..32769. This actually makes more sense to me.

I agree but you can't believe how may LR users hate it and want the same feedback as Photoshop. I think LR should give them that option but I prefer the current behavior. For new users, it makes more sense and is far easier for them to grasp too.

[Jim Kasson]

With floating point notation, the number of encoded colors can be arbitrarily large.

<nit-pick mode on> The number of discrete values expressible as an n-bit unsigned integer and as an n-bit floating point number are equal for all values of n in the set of positive integers, are they not? In both cases, the number of discrete values is 2^n. Actually, as I think about it, the number of discrete values expressible in floating point notation is somewhat less, since some bit combinations are not allowed, for instance if subnormal numbers aren't allowed. Then there's signed zero.<nit-pick mode off>

Good post, though. And I think Andrew is on the right track.

Jim

[Tony Jay]

So based on the comments on color gamut vs. color space, how does this sound? I've obviously done some picking pieces from here and there to come up with this. It's got nothing to do with a video per se, but more of a distilling of the last number of pages on color gamut vs. color numbers:

IMHO an excellent summary that really does highlight both the similarities and differences between colourspaces.
Perhaps, the only possible omission may be a little more detail on colour perception with regard to distinguishing between two colours as being different. The ability of the eye distinguish two colours as different is more limited in the yellows but is better in the greens and blues. I fully agree that the issue is a secondary one in the discussion of colourspaces but it may have a place as a way of building insight particularly with regard to actually editing, softproofing, and printing images.

Best regards

Tony Jay

[fdisilvestro]

<nit-pick mode on> The number of discrete values expressible as an n-bit unsigned integer and as an n-bit floating point number are equal for all values of n in the set of positive integers, are they not? In both cases, the number of discrete values is 2^n. Actually, as I think about it, the number of discrete values expressible in floating point notation is somewhat less, since some bit combinations are not allowed, for instance if subnormal numbers aren't allowed. Then there's signed zero.<nit-pick mode off>


Jim

Would it be right to say that the difference is that integer values are equally spaced while floating point are not?

[digitaldog]

I could add something like: A deltaE of less than 1 between two colors is said to be imperceptible but to complicate matters, there are several formulas for calculating this metric. Further the ability of the eye distinguish two colors as different and is more limited for yellows but is better for greens and blues. This just adds even more difficulty in assigning a meaningful and accurate number of colors to these colors spaces.

[GWGill]

For the purpose of this debate (which wasn't meant to be a geeks' orgy, btw), this makes as much sense as saying to someone who asked you how far away is the nearest town: "Sorry, Sir, that's a nonsense question. Everybody knows that distances are conceptually continuous, therefore it's wrong to talk about number of miles or kilometers."

You have taken what I wrote and now misconstrued it into something I didn't say. Counting the "number of colors" in a colorspace is not the only way to measure it. By all means measure it in some device independent way like delta E. But introducing such a metric is hardly an easy explain to color novices.