Color management myths and misinformation video

[digitaldog]

Due to the recent and unfortunate onslaught of video’s about sRGB vs. Adobe RGB (1998), I decided to create a video addressing color management myths and misinformation. Below is a basic outline for slides and discussion. I’d like feedback from the group, especially those that had a say in the last and lengthy discussion here on the subject.

What’s missing? What’s wrong with the statements? What topics need to be drilled down farther if any?

Adobe RGB (1998) produces duller colors.
Not so, only if you treat Adobe RGB (1998) as sRGB. Plan to show the opposite, what happens if you treat Adobe RGB (1998) as sRGB. Use Assign Profile command to properly illustrate this. I’d like to show that treating one as the other isn’t the fault of the color space but rather the user! Analogy would be resampling a high rez file down to a tiny image using Nearest Neighbor and showing the results. It isn’t the resample algorithm at fault, it’s the user (don’t do dumb things).

There are no sRGB printers. There are printers that expect sRGB for conversion to the native output color space. Explain the only device that could produce sRGB or Adobe RGB (1998) is an emissive display and these are theoretical. Like to reference the 1998 post by Michael Stokes of HP. Show gamut maps of Noritsu, Lightjet etc compared to sRGB. Discuss an Epson printer with it's driver that expects RGB, not CMYK data, yet printer isn't an RGB printer.

Dismiss the idea “there are no printers with a color space (aka output space) that is larger (holding more volume of data) than sRGB”. Will reference my Gamut video but would show 3D gamut maps from a few devices compared to sRGB.

Adobe RGB (1998) has more colors than sRGB. Discuss gamut (range of colors) vs. number of colors (encoding). I’d like to discuss Gamut Volume as reported in ColorThink but not go too deep (it’s geeky). This is one area I could use more feedback. How deep to go, maybe some discussions about Gamut Volume in general. Bottom line, saying Adobe RGB (1998) has more colors isn’t as accurate as saying Adobe RGB (1998) has a larger gamut volume.

There is nothing wrong with sRGB or Adobe RGB (1998), There is not and should be no debate about which is better! Depending on the use of the image, one is more appropriate (will give better results) than the other.

Comments?

Thanks.

[Slobodan Blagojevic]

Andrew, most of those are straw-man arguments. The only one that clearly isn't is #3.

#4 is interesting, because it is confusing, including myself. As someone noticed, if we perceive more differentiated colors (or variations, nuances of, more saturated) in say Adobe RGB, than in plain English it does have more colors.

[sandymc]

Andrew,

With all due respect, you're going down the same useless rat hole that resulted in a 27 page thread by talking about sRGB vs Adobe. As I tried to point out in my single post, it's the wrong question; a video that addresses the right question (color managed work flow vs. non-color managed workflow and how the difference can result in people believing that one color space is somehow "better") might be useful. Anything peppered with sRGB and Adobe RGB - not.

Sandy

[digitaldog]

Andrew, most of those are straw-man arguments. The only one that clearly isn't is #3.

Yes they are, that's why the video is to address myths and misinformation.

With all due respect, you're going down the same useless rat hole that resulted in a 27 page thread by talking about sRGB vs Adobe.

Yes I am, that's kind of the point of the video, a rebuttal if you will, much of it discussed here but in one location for those who don't want to read through 27 pages. A video too would allow the use of 3D gamut plots, showing the affect of the Assign Profile command and so forth. IOW, I could write an article or just copy and paste all the good stuff from those 27 pages, but a 10-15 minute video would IMHO be much more effective to present to someone who is looking for the crux of the facts. It isn't intended for this audience.

[D Fosse]

I'd have to agree with Sandy. A more useful approach is "what happens when color management breaks down" - and the various ways in which that can happen. Explain how and why it's essential to use a profile that actually describes the color space it refers to. "There are no sRGB printers" would fit nicely in here. Using the wrong display profile also.

Discussing the relative merits of different color spaces is indeed the wrong question and beside the point.

An important point to get across, IMO, is that color management isn't nearly as difficult as many try to make it. The problems start when color management breaks down and stops.

[digitaldog]

I'd have to agree with Sandy. A more useful approach is "what happens when color management breaks down" - and the various ways in which that can happen. Explain how and why it's essential to use a profile that actually describes the color space it refers to. "There are no sRGB printers" would fit nicely in here. Using the wrong display profile also.

I was hoping the Assign Profile demo would illustrate this. If you take Adobe RGB (1998) data and Assign sRGB, what you'd see on-screen would appear incorrect despite the numbers not changing and I'd have the Info palette up to show this. This illustrates that the numbers must have the correct scale associated or they "look dull" or "look too saturated". It is a color management failure. If you move forward with that incorrect tag, the output will of course suffer. So the numbers are and have been correct from the beginning. The color management failure is the improper assignment (scale) which affects everything moving forward. Message to viewer: don't do that!

Discussing the relative merits of different color spaces is indeed the wrong question and beside the point.

Well there are merits to posting sRGB for web viewing right? And there are merits to sending Adobe RGB (1998) instead of sRGB to the printer as Bill illustrated in his flower image. What the message should be IMHO is, test both (here's how). Just as Bill did. Take Adobe RGB (1998) and convert to sRGB, maybe the same with ProPhoto too. Take each and properly send them to a printer and examine the results.

An important point to get across, IMO, is that color management isn't nearly as difficult as many try to make it. The problems start when color management breaks down and stops.

Duly noted and I agree. That's why I suggest that assigning sRGB to Adobe RGB data or vise versa is wrong but not due to the color spaces but the user we handled the data incorrectly.

[Slobodan Blagojevic]

Yes they are, that's why the video is to address myths and misinformation...

I am sure you understand the meaning of "straw-man" argument, i.e., it is you who first creates a "myth," and then gloriously debunks it. Otherwise, nobody claims that "Adobe produces duller colors," nobody claims there are "sRGB printers," etc., these are just your hairsplitting, literal interpretations of contextual shorthands people use.

P.S. I am saying the above with friendly intentions, not to provoke a rhetorical fight

[milt]

I've read only snippets here and there from the long post, getting bored quite quickly.  Your stamina on this issue is remarkable Andrew!

If I understand your target audience for the video, I think I'm with Slobodan on point #4.  The conceptual difference between "number of colors" and "gamut volume" seems too geeky.

Milt

[mrenters]

Andrew, most of those are straw-man arguments. The only one that clearly isn't is #3.

#4 is interesting, because it is confusing, including myself. As someone noticed, if we perceive more differentiated colors (or variations, nuances of, more saturated) in say Adobe RGB, than in plain English it does have more colors.

I agree that saying the number of colours between sRGB and AdobeRGB is the same is confusing.  At 8 bits per channel, both spaces have exactly 16,777,216 individually addressable colours. Some of these may be <1dE apart and therefore indistinguishable from each other in human vision. The range of colours that each of those 16,777,216 RGB combinations represent is certainly larger in AdobeRGB than sRGB because of AdobeRGB's larger gamut and more saturated colours. If you think of each of those combinations as paint chips at the paint store, then the AdobeRGB paint company offers you more "colours" to pick from when you want to repaint your room.

Martin

[MarkM]

One thing that people including Gary butcher is the difference between clipping and profile mismatches. Gary kept talking about the dulling of colors that occurs when you interpret AbobeRGB numbers as if they were sRGB, but he referred to it on several occasions as clipping, which it's not. The difference would be pretty easy to show visually with a photo that exhibits the classic blocked up areas of clipping vs hue and saturation shifts of a profile mismatch.

Regarding number 4, I think a good, accurate analogy might help people understand. I think the main problem is that the question is ill-formed. AdobeRGB and sRGB are just spaces, they don't inherently have any information (other than specifications for primaries, white point, etc). Until you actually have a pixel, there isn't any color information. The analogy I like, which I think is accurate enough, is a thermometer. If you have a thermometer that goes from 0-100º C and another that goes from 0-200º C, which one has more information? It's kind of a dumb question right? Thermometers don't have information — the information is the actual measurements. If I take to two measurements with each, do I have more information if I use the 0-200º thermometer? No, you have exactly two measurements worth of information in each case. Those measurement could be specified by how accurate they and this could be expressed as bits. You probably have to draw the wonk-line somewhere before you start talking about Claude Shannon, but I'm not sure where that is. A question you can ask with this analogy is this: how fine of a distinction in temperature can a person make? If you determined that and added a scale in these units to the thermometer, then the 0-200º thermometer would measure more of these units. But that's different than saying it has more information and very different from saying that's all the precision we need in practice.

[digitaldog]

I am sure you understand the meaning of "straw-man" argument, i.e., it is you who first creates a "myth," and then gloriously debunks it. Otherwise, nobody claims that "Adobe produces duller colors," nobody claims there are "sRGB printers," etc., these are just your hairsplitting, literal interpretations of contextual shorthands people use.

I'm glad you brought that up because after you wrote the first piece, I did look it up (I've heard it for years but didn't understand your context). One example I saw:

Description of Straw Man
The Straw Man fallacy is committed when a person simply ignores a person's actual position and substitutes a distorted, exaggerated or misrepresented version of that position. This sort of "reasoning" has the following pattern:
Person A has position X.
Person B presents position Y (which is a distorted version of X).
Person B attacks position Y.
Therefore X is false/incorrect/flawed.
This sort of "reasoning" is fallacious because attacking a distorted version of a position simply does not constitute an attack on the position itself. One might as well expect an attack on a poor drawing of a person to hurt the person.

So someone watches a video where a person talks about duller colors and sRGB printers. What can and should be done to dismiss those fallacies and reeducate that viewer? Is this idea of mine pointless (I'm willing to abandon it, but I think folks need some facts presented to them properly).

[Alan Goldhammer]

I'd have to agree with Sandy. A more useful approach is "what happens when color management breaks down" - and the various ways in which that can happen. Explain how and why it's essential to use a profile that actually describes the color space it refers to. "There are no sRGB printers" would fit nicely in here. Using the wrong display profile also.

Discussing the relative merits of different color spaces is indeed the wrong question and beside the point.

An important point to get across, IMO, is that color management isn't nearly as difficult as many try to make it. The problems start when color management breaks down and stops.

I agree and think also that this is a bit of a fools errand.  What is the audience for this video, not those of us on LuLa who have learned all of this.  I don't think it's the general public that shoots JPEG and posts to various media sites.  Jeff Schewe condenses all of this to a relatively few very well written pages in his books on the digital negative and print.  Your book can be gone to for more in depth coverage (and I have a well worn copy!).  Anyone who does printing should have some good knowledge of things, those who just post to the Internet could care less.

[digitaldog]

Regarding number 4, I think a good, accurate analogy might help people understand. I think the main problem is that the question is ill-formed. AdobeRGB and sRGB are just spaces, they don't inherently have any information (other than specifications for primaries, white point, etc). Until you actually have a pixel, there isn't any color information. The analogy I like, which I think is accurate enough, is a thermometer. If you have a thermometer that goes from 0-100º C and another that goes from 0-200º C, which one has more information? It's kind of a dumb question right? Thermometers don't have information — the information is the actual measurements. If I take to two measurements with each, do I have more information if I use the 0-200º thermometer? No, you have exactly two measurements worth of information in each case. Those measurement could be specified by how accurate they and this could be expressed as bits. You probably have to draw the wonk-line somewhere before you start talking about Claude Shannon, but I'm not sure where that is. A question you can ask with this analogy is this: how fine of a distinction in temperature can a person make? If you determined that and added a scale in these units to the thermometer, then the 0-200º thermometer would measure more of these units. But that's different than saying it has more information and very different from saying that's all the precision we need in practice.

Absolutely brilliant! Great analogy. I'm going to want to steal it <g>.
In terms of how fine a distinction with respect to this topic, it has to be one that is visible (because if you can't see it, it's not a color).
IF nothing comes about further, the analog above is lovely and well worth the post.

[Slobodan Blagojevic]

... do I have more information if I use the 0-200º thermometer?...

You do if your temperature is 150 degrees.

[MarkM]

You do if your temperature is 150 degrees.

That's an interesting and subtle point. I'm not convinced it's true, though. I think it would be true to say you have more accurate information. If you take the measurement of 150º with the 0-100º thermometer and it reads 100º, you've clipped, but you still have one measurement worth of (inaccurate) information. I fully admit that I'm not an expert on information theory, so I could be way off here.

[digitaldog]

You do if your temperature is 150 degrees.

Well yes and you're dead  
But then this goes back to the question of how fine a distinction. If indeed the difference is less than a dE of one, we can't see it, it isn't a color.
In terms of a thermometer that has a range, and the task is to see if you have a fever, is it necessary to have one that goes 0-200? It doesn't apply in the context of measuring your temperature unless the goal is to measure a dead person while being cremated.

[digitaldog]

I agree and think also that this is a bit of a fools errand.

You might indeed be correct.

What is the audience for this video, not those of us on LuLa who have learned all of this.

No, not this audience by a long shot.
The last 1 minute video I did was aimed at someone who doesn't have a clue what the differences are between sRGB and Adobe RGB to show that it's really easy to say "just use sRGB". Then there is reference to a long geeky video on color gamut. I'm thinking something in-between and one that as the title suggests talks about color management myths. But maybe it is a fools errand. I'm totally open to that possibility. I sure wish there was a way to rebut some of the silliness out there but not make it a complex treatise on color management. And in that respect, the number of colors vs. gamut volume seems out of place despite what seems to be a fantastic analogy from Mark.

[Torbjörn Tapani]

It would be better with a video explaining color management in so clear and concise way that there are no more misconceptions. Such a video could have an audience other than this preaching to the choir stuff.

Others have said, loosely quoted "as simple as possible but no simpler" and "increasing the signal".

[digitaldog]

It would be better with a video explaining color management in so clear and concise way that there are no more misconceptions. Such a video could have an audience other than this preaching to the choir stuff.
Others have said, loosely quoted "as simple as possible but no simpler" and "increasing the signal".

I'm beginning to think you folks are correct, this is a fools errand. When I ask people for advise, I generally take it (depending on the people of course  ).
The thread was useful if for anything, the thermometer analogy. Thanks!

[Tony Jay]

Andrew, in my humble opinion the issue has never been about an argument pitting sRGB and AdobeRGB against each other.

Instead Gary was committing acts of violence against a rational colour management-based workflow.
Far better to explain the place of the various colourspaces in a rational workflow.
How, depending on one's goals at shooting and the intended output, one can make very rational choices about assigning workspaces in camera (not applicable when shooting RAW - this should be explained too).
Some time should be spent discussing the rationale behind Adobe giving photographers no choice in their working colourspace - it is ProPhotoRGB for everyone - so many of the strange and incorrect misconceptions that Gary has can be put to bed here.
A good explanation of colourspaces on input, working colourspaces, and output colourspaces integrated into the above discussion.
How it is possible to take an image from one colourspace to another without doing the sort of violence to it that Gary demonstrated i.e. what happens with a colourspace conversion and the place of softproofing.

Thinking about it - I am not sure that any of us could adequately explain any of this in few minutes but the challenge is to make rational colour management-based workflows accessible to the very owners of "Digital Rebels" that Gary Fong believes cannot understand principles of colour management.
You may note that I have not included Gary's classification of individuals who are not interested in colour management - they will no more view Gary Fong's videos on colour management than they will Andrew Rodney's.
I was once one of those Digital Rebel owners who knew nothing about colour management but gradually became dimly aware that it may actually have some relevance - the journey started there.

So, I strongly believe that we should not get sucked into Gary Fong's misinformation agenda (the sRGB vs AdobeRGB non-issue) and concentrate on explaining rational and practical colour management-based workflow decision-making.

My 0.02c

Tony Jay