Pages: 1 ... 7 8 [9] 10 11 ... 24   Go Down

Author Topic: Color management myths and misinformation video  (Read 84494 times)

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Re: Color management myths and misinformation video
« Reply #160 on: August 31, 2014, 01:29:28 pm »

Here's another image that's illuminating when opened in ColorThink. This image is a 256 x256 slice of the RGB space where the blue value is zero. ColorThink seems to limit its color list to 10000 points and it seems to decide this by tossing values at set RGB value intervals (culling more in the shadows and fewer in the highlights).   I think it's safe to assume the color list is not a good indication of the number of discernible colors in a file.

Mark,

Your Tiff image did not display but was downloaded instead. If you want to avoid JPEG, then I would suggest PNG which I used to repost this image.

Bill
Logged

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Color management myths and misinformation video
« Reply #161 on: August 31, 2014, 01:39:20 pm »

It might also be nice to post the link to the X-Rite/Pantone Hue Color Test that viewers can take.  This really highlights how we all see and perceive color.
Well kind of. I think it's fun, it does give people an idea of what the correct vision test could do, but it's largely a marketing vehicle. Considering how it arrives to a viewer (on the internet), that depending on browser, calibration of the display and other factors, not sure. I could just as easily create two squares in Photoshop to show on a video that had a dE of less than one, 1 and higher but the same viewing caveats would apply. Maybe a document one could download and view in Photoshop? Or just the recipe for the colors one could create like I did to test CT?
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

MarkM

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 428
    • Alaska Photographer Mark Meyer
Re: Color management myths and misinformation video
« Reply #162 on: August 31, 2014, 04:00:52 pm »

Since we appear to agree that the number of colors have to be discernible to be valid…

This question has been interesting to think about and I think it's quite challenging. Having said that, I don't think I agree with this.

I don't think the question is valid. Everything in CIE Colorimetry is based on continuous functions. Although the original color matching data is discreet, and any spectral measurements you get from a spectrometer are also discrete, the first thing we do is interpolate to smooth functions. When you look at the formulas for moving from spectral data to a tristimulus space, they are full of integrals, nots sums. It's continuous functions from the ground up. So asking about discrete data requires you to sample from the model, but colorimetry doesn't give you good tools to make these samples for the current problem.

Additionally, although we often casually talk about XYZ or RGB numbers as being colors, that's not entirely true. Colorimtery traffics in color stimuli not colors. And it only describes how different stimuli under very specific conditions match. Colorimetry is not a system for identifying color perceptions.

While many people in this thread have pointed out that different stimuli can result in the same color perception, the opposite is also true. For example, consider the sRGB value (130, 70, 15). Does that stimulus map to one perception? If you were to count the number of distinct colors in the following image, will that RGB value be tallied as one perception? If so, which perception does it map to, the top center square or the bottom (they're both the same RGB value)?


(from: http://www.lottolab.org/articles/illusionsoflight.asp)

We also have difficult semantic problems. We might be happy to define colors based, by definition, on human perception, but we freely use the term color for any perception. For example some birds are known to be tetrachromats with an additional cone sensitive to ultraviolet light. This means that stimuli that would be identical perceptions to us are different colors to them. We're happy to extend the definition of colors to distinctions birds can make and simply say that these are two colors that we can't distinguish, but something else can. We frequently talk about perceptions that we can't perceive, but that something can or that some tool can measure. For example sounds that are below our frequency threshold that elephants use to communicate or smells that only by dog can smell. So if we have color stimuli that we can't distinguish, but which the camera can, isn't it just easier to call these different colors, but with differences below our threshold?

There are also practical problems. If you look at the literature, the attempts to define the number of colors are all over the map. Edward Titchener came up with 33,000 in 1896, Edwin Boring came up with about 10 times that number around the same time. Deane Judd estimated 10 million while David MacAdam estimated 17,000 — both legendary color scientists. Then you have Mark Fairchild going out on a limb and claiming an infinite number: http://www.rit-mcsl.org/fairchild/WhyIsColor/files/ExamplePage.pdf. Clearly there's some disagreement about how to approach the question.

All of these problems go away, if we just call colorimetry what it is — a model – and avoid questions that ask this model things it wasn't designed for. Questions it handles nicely are things like, what is the range of stimuli, or how many color values are in a file, or even how many cubic ∆Es does a space contain — these are are very clean and easy. Questions like, how many distinct perceptible colors are in a working space, don't easily fit with the model and therefor become messy and confusing and identifying the reason for the messiness is probably more valuable than any answer you could come up with to the original question.
« Last Edit: August 31, 2014, 10:23:39 pm by MarkM »
Logged

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Color management myths and misinformation video
« Reply #163 on: August 31, 2014, 04:12:58 pm »

While many people in this thread have pointed out that different stimuli can result in the same color perception, the opposite is also true. An example might help. Consider the sRGB value (130, 70, 15). Does that stimulus map to one perception? If you were to count the number of distinct colors in the following image, will that RGB value be tallied as one perception? If so, which perception does it map to, the top center square or the bottom?
Lot's to chew on with that post Mark. It seems to enforce even more the reason not to say: Adobe RGB (1998) has more colors than sRGB. If someone wants to know why, we've got several pages of reasons.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Slobodan Blagojevic

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 18090
  • When everyone thinks the same, nobody thinks
    • My website
Re: Color management myths and misinformation video
« Reply #164 on: August 31, 2014, 05:30:44 pm »

... the reason not to say: Adobe RGB (1998) has more colors than sRGB. If someone wants to know why, we've got several pages of reasons.

Ok... so they have the same number of colors... what's different then? And btw, try to answer in one, simple-English sentence, not "several pages."

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Re: Color management myths and misinformation video
« Reply #165 on: August 31, 2014, 05:32:43 pm »

Lot's to chew on with that post Mark. It seems to enforce even more the reason not to say: Adobe RGB (1998) has more colors than sRGB. If someone wants to know why, we've got several pages of reasons.

This thread and the prior closed down thread have given us a lot to think about and has been very instructive to many of us and hopefully some of these topics will make it into your movie, should you (hopefully) decide to go forward with it.

Debunking Mr. Fong's mis-statements would likely lead down into a rat hole, but a movie illustrating and explaining what has been discussed would be helpful. What happens when basic color management is not used?

If the colors of a scene fit into sRGB or Adobe RGB, there is nothing wrong with those color spaces. However, as has been discussed, many real world scenes won't fit into either of these spaces without clipping and it is here that ProPhotoRGB is useful. I'm really not interested in how many colors this space can contain but rather that it can encode saturated colors often found in nature. As discussed, the digital camera does not have a gamut in the strict sense, but it does record a wide range of colors. To illustrate, this Dahlia was photographed and rendered into ProPhotoRGB with ACR using the Adobe Standard profile with normal settings with PV2012 and no saturation of vibrance adjustments. It occupies nearly the full gamut of ProPhotoRGB as shown in the preview (uploaded in Adobe RGB and likely problematic with a non-color managed browser). Severe saturation clipping occurs when rendering into sRGB and there is an easily noted loss of saturation when viewed on a wide gamut monitor.

The ProPhoto image can't be printed with current technology, but it makes sense to capture as much gamut as possible. I would not recommend printing this image in sRGB (sending sRGB to the printer).

Bill

« Last Edit: August 31, 2014, 05:38:53 pm by bjanes »
Logged

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Color management myths and misinformation video
« Reply #166 on: August 31, 2014, 05:34:00 pm »

Ok... so they have the same number of colors... what's different then? And btw, try to answer in one, simple-English sentence, not "several pages."
What's different? The range of colors is different. G255 in sRGB is in a different location within human vision as defined by the CIE chromaticity plot than Adobe RGB (1998).
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Color management myths and misinformation video
« Reply #167 on: August 31, 2014, 05:39:03 pm »

It's not just color management that fills the web with urban legend, just read this:
Quote
Raw does have a greater dynamic range. JPG's are 8 bit per channel.
Let's not go there  ;D
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Slobodan Blagojevic

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 18090
  • When everyone thinks the same, nobody thinks
    • My website
Re: Color management myths and misinformation video
« Reply #168 on: August 31, 2014, 05:57:15 pm »

... G255 in sRGB is in a different location within human vision as defined by the CIE chromaticity plot than Adobe RGB (1998).

Let me see if I get it: are you saying that both spaces have a color defined as G255 (by which I assume you mean a maximum pure green?), just in different locations? Are those the two circles I pointed out in the attached image (assume that I more or less correctly pinpointed the locations)?

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Color management myths and misinformation video
« Reply #169 on: August 31, 2014, 06:01:51 pm »

Let me see if I get it: are you saying that both spaces have a color defined as G255 (by which I assume you mean a maximum pure green?), just in different locations? Are those the two circles I pointed out in the attached image (assume that I more or less correctly pinpointed the locations)?

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.
« Last Edit: August 31, 2014, 07:00:09 pm by digitaldog »
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: Color management myths and misinformation video
« Reply #170 on: August 31, 2014, 06:42:41 pm »

The plot you provided, that Gary can't understand or use, shows this quite clearly. That horseshoe shape represents human vision.

If we're going to use chromaticity diagrams, can we at least use u'v' instead of xy, which unrealistically overemphasizes the importance of green changes? u'v' isn't perceptually uniform, but it's a lot better than xy. Near as I can tell, the only advantage of xy is that it's easier to calculate on the back of a napkin. If there's a computer nearby that advantage goes away.

One thing that's nice about xy is that colors additively mix along straight lines, so emissive display gamuts are triangles. That's true in u'v' as well.

There was a 1958 Farnsworth paper that had a chromaticity space that turned the MacAdam ellipses into circles, but I can't find it.

Jim

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Color management myths and misinformation video
« Reply #171 on: August 31, 2014, 06:54:28 pm »

If we're going to use chromaticity diagrams, can we at least use u'v' instead of xy, which unrealistically overemphasizes the importance of green changes? u'v' isn't perceptually uniform, but it's a lot better than xy.
We (Slobodan) could but considering the audience and the lesson being taught, the method he used is certainly adequate.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: Color management myths and misinformation video
« Reply #172 on: August 31, 2014, 06:56:03 pm »

We (Slobodan) could but considering the audience and the lesson being taught, the method he used is certainly adequate.

Yes, I was sloppy there. I meant in the video.

Sorry, Slobodan.

Jim

Tony Jay

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2965
Re: Color management myths and misinformation video
« Reply #173 on: August 31, 2014, 07:21:43 pm »

Let me see if I get it: are you saying that both spaces have a color defined as G255 (by which I assume you mean a maximum pure green?), just in different locations? Are those the two circles I pointed out in the attached image (assume that I more or less correctly pinpointed the locations)?
Maybe to point out the obvious Slobodan (apologies if that is the case) but 0,255,0 in sRGB can likely be represented by some other set of numbers in AdobeRGB say 0,221,0.
To achieve this though requires a colourspace conversion.
Not doing that gives the rather unfortunate results demonstrated by you know who where the red in one colourspace became some strange colour in another.
Obviously, the actual colour represented by 0,255,0 in AdobeRGB cannot be directly represented in sRGB.
How one deals with this scenario is dealt with by rendering intents when a colourspace conversion is done from AdobeRGB to sRGB.

Tony Jay
Logged

GWGill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 608
  • Author of ArgyllCMS & ArgyllPRO ColorMeter
    • ArgyllCMS
Re: Color management myths and misinformation video
« Reply #174 on: August 31, 2014, 09:47:13 pm »

Saying "wider gamut" just shifts the burden to defining then "gamut." Besides, "more vivid colors" simply means more colors.
That's the point. Shift the burden to a valid viewpoint, rather than a nonsense one.

Colorspaces are conceptually continuous, not discrete, therefore it's wrong to talk about number of colors.
In practice they are often quantized for compact transmission of images, but this is orthogonal to gamut size, and rather arbitrary (ie. a power of 2). Ideally such quantization is too small to be perceivable and so isn't something you use to make the colors countable.
Logged

MHMG

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1285
Re: Color management myths and misinformation video
« Reply #175 on: August 31, 2014, 09:59:13 pm »

Well, maybe we did overthink some of the points in this discussion, but hopefully it will help Andrew refine the way he presents some of the concepts in his upcoming video :). For me, it has been fun.  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


« Last Edit: August 31, 2014, 10:05:01 pm by MHMG »
Logged

MarkM

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 428
    • Alaska Photographer Mark Meyer
Re: Color management myths and misinformation video
« Reply #176 on: August 31, 2014, 10:11:58 pm »

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.

I'm not quite sure what you mean by this one. You can easily find pairs of colors in an RGB space that will be very close to each other like the previously-mentioned cyan pair that are separated by only .02 ∆E. I have trouble imagining a situation in which you could discern between those two values, but I'm not sure how our ability to discern color differences changes in different situations. Also spaces like ProPhoto RGB have colors that fall outside the spectral locus — they aren't real colors. I think it's hard to argue that they are visually discernible in any meaningful way.

Logged

MHMG

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1285
Re: Color management myths and misinformation video
« Reply #177 on: September 01, 2014, 08:49:17 am »

I'm not quite sure what you mean by this one. You can easily find pairs of colors in an RGB space that will be very close to each other like the previously-mentioned cyan pair that are separated by only .02 ∆E.

Yes, one can choose some pairs of values that will not be discernible when placed side-by-side each other but each one placed against another appropriate surround condition is discernible. Hence, has equal importance to construction of a complex image geometry.  In other words even two values with .02 ∆E differences are equally useful addressable values to be used when filling in all the pixels needed to build up a real image. And especially when paring them in situations where the pair falls below 1 JND, they are necessary to build shallow smooth gradients. If we throw out all combinations of addressable colors that fall below 1 JND in a side-by-side viewing condition, we'd have to resort to much more aggressive dithering in the reproduction to mimic full continuous tone gradients of tone and color.


I have trouble imagining a situation in which you could discern between those two values, but I'm not sure how our ability to discern color differences changes in different situations. Also spaces like ProPhoto RGB have colors that fall outside the spectral locus — they aren't real colors. I think it's hard to argue that they are visually discernible in any meaningful way.


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. What happens all too many times is that clipped colors bunch up and consequently start to present large areas of color and tone that don't hold the proper visual contrast relationship. Even if they are distinguishable from other nearby colors the relationships between the colors don't look natural.  Hence, all out of gamut colors, whether they are real or imaginary ones forced into gamut can become very problematic for image quality. It's why we spend so much effort dealing with out of gamut colors and tones when trying to make a natural looking reproduction of a naturally occurring scene. For any photographer that likes hyper saturated colors, Prophoto can be a dangerous place to do image editing!

After posting last night, I  thought I should perhaps have clarified another point, namely, when any given RGB triplet is assigned to another color space (i.e the Gary Fong misrepresented demonstration) only pure black (0,0,0) and pure white (255,255,255) retain the exact same assigned L*, a*, and b* values in the new color space. So, for all other triplets, LAB values may be assigned that noticeably affect hue, and lightness properties not just chroma properties as well. Nevertheless, chroma is only color property where the numerical range (minimum to maximum chroma values) must be expanded to accommodate the transformed color set when moving from "smaller" to "bigger" color spaces. The range of lightness values and the range of hue values present in each of the working RGB color spaces remains the same. Hence, the best way to describe the merits of different color spaces to non technical folks without introducing technically weak analogies may simply be to say that because we are dealing with digitally encoded versions of any color space and thus the assignable values represent a finite total number of addressable color values, the different RGB working color spaces are not needed in order to accommodate a larger number of discreet colors. They are needed to accommodate specific colors that have greater chroma values as a fundamental aspect of their three visually perceivable properties, ie. lightness, hue, and chroma.
 
cheers,
mark
http://www.aardenburg-imaging.com
« Last Edit: September 01, 2014, 08:53:19 am by MHMG »
Logged

fdisilvestro

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1852
    • Frank Disilvestro
Re: Color management myths and misinformation video
« Reply #178 on: September 01, 2014, 10:12:55 am »

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.).

While this is true, I think it is not the right approach. If we consider the following statement (with which I fully agree):

Colorspaces are conceptually continuous, not discrete, therefore it's wrong to talk about number of colors.
In practice they are often quantized for compact transmission of images, but this is orthogonal to gamut size, and rather arbitrary (ie. a power of 2). Ideally such quantization is too small to be perceivable and so isn't something you use to make the colors countable.

The last phrase says it all, the idea of quantization (usually related to bit depth) is to have it so small that is not perceivable. The fact that we work with 8, 16, 32 bits is a practical one, due to the binary system commonly used.

In other words, use the number of bits necessary to make the quantization step non-discernible.

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.

If they were visually discernible then you need to reduce the quantization step (or add bit depth), that's the purpose of it. In practice noise and the limitations of the visual system set the limit to required bit depth.

Citing the example of the 2 cyan patches that Mark provided, in 8 bits they are separated by a DeltaE of approximately 0.02. If they still are discernible then move to 16 bits, where there will be 255 patches between them, still discernible? (I can bet they will be not , not matter what condition) move on, 32, 64, 128 you name it. Actually, you don't need that many, consider the document mentioned here:

Then you have Mark Fairchild going out on a limb and claiming an infinite number: http://www.rit-mcsl.org/fairchild/WhyIsColor/files/ExamplePage.pdf.

If you read it, "infinity" is just 18*10^33, which translated to base 2 results in 2^114. If we use 3 channels such as RGB then we just need 114/3 = 38 bits per channel to address all those colors.

Really, forget about the numbers.

Slobodan Blagojevic

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 18090
  • When everyone thinks the same, nobody thinks
    • My website
Re: Color management myths and misinformation video
« Reply #179 on: September 01, 2014, 11:25:49 am »

... Shift the burden to a valid viewpoint, rather than a nonsense one.

Colorspaces are conceptually continuous, not discrete, therefore it's wrong to talk about number of colors...

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."
Pages: 1 ... 7 8 [9] 10 11 ... 24   Go Up