Dont know if this is going to cause a long debate or not. Im soon going to be calibrating my mits wd73835 which has full cms controls with a chroma 5 meter and calman standard. Ive been doing alot of reading and cant determine which setting i should use in calman for delta e. Do i use 1976 or 1994 and why? Thanks.

Mike

"Don't know if this is going to cause a long debate or not"

Likely it will - especially if one were to add CIE2000 to the debate. Check this thread out:

http://www.avsforum.com/avs-vb/showthread.php?t=944837

No one that I can find working in color science would support staying with the old measures of color error (and the weak perceptual uniformity of the color spaces they're based on) yet the video industry seems intransigent.

For grayscale the consensus seems to be using dE76. But for Gamut/CMS using dE94 gives us the advantage because from dE94 you can get dL, dC and dH. So when you have a large error with dE94 you can see where the error is coming from is it Luminance, Chroma or Hue. We have tested with dE2000 but have not found it to be useful it was included as an option but I have never found a reason to use it.

On what basis did you determine usefulness?

CIE94 and CIELAB are almost identical for grayscale. CIELUV scales a little higher, but is otherwise similar. CIEDE2000 has not been widely adopted outside of the textile industry.

The more modern dE formulas begin to show real differences from the 1976 formulas when you include the lightness component, which you do not for grayscale.

CIE94 and CIELAB are almost identical for grayscale. CIELUV scales a little higher, but is otherwise similar. CIEDE2000 has not been widely adopted outside of the textile industry.

The more modern dE formulas begin to show real differences from the 1976 formulas when you include the lightness component, which you do not for grayscale.

So would it make sense to use 76 for greyscale and 94 for color calibration as stated in another reply?

Mike

So would it make sense to use 76 for greyscale and 94 for color calibration as stated in another reply?

Mike

Yes and that is what we recommend for use within CalMAN.

I'm actually quite interested in this. Recently, I've been doing a bit of experimenting in CalMAN.

For colour I disabled gamma correction, as you can't make meaningful colour adjustments with it turned on. As everyone seems to recommend, dE'94 broken out into the LCH components seems to work best.

For greyscale, I then found that dE u'v' worked best. I don't have any control over gamma so it's not relevant to greyscale performance in my case—I only want to see the colour error, and dE u'v' is the only one that takes luminance out of the equation with gamma correction disabled.

It also seems stricter than other dE calculations for greyscale. I've been using the simulated sensor for some of this testing, which can produce quite erratic results at each level. What I have seen a few times is errors where, for example 50 and 60 IRE are roughly the same error level in most methods of calculation, but one of them may be higher than the other in dE u'v'. e.g dE 3 for both with most calculations, but dE 3 and 4 in u'v'.

Is there anything wrong with this method?

I'm actually quite interested in this. Recently, I've been doing a bit of experimenting in CalMAN.

For colour I disabled gamma correction, as you can't make meaningful colour adjustments with it turned on. As everyone seems to recommend, dE'94 broken out into the LCH components seems to work best.

For greyscale, I then found that dE u'v' worked best. I don't have any control over gamma so it's not relevant to greyscale performance in my case—I only want to see the colour error, and dE u'v' is the only one that takes luminance out of the equation with gamma correction disabled.

It also seems stricter than other dE calculations for greyscale. I've been using the simulated sensor for some of this testing, which can produce quite erratic results at each level. What I have seen a few times is errors where, for example 50 and 60 IRE are roughly the same error level in most methods of calculation, but one of them may be higher than the other in dE u'v'. e.g dE 3 for both with most calculations, but dE 3 and 4 in u'v'.

Is there anything wrong with this method?

What setting disables gamma correction? Im assuming its the box that says correction under the delta e tab? Thanks.

Mike

What setting disables gamma correction? Im assuming its the box that says correction under the delta e tab? Thanks.
Options > DeltaE > Gamma Correction.

This must be disabled when making colour adjustments/measurements in CalMAN. You should be warned that it doesn't seem to stick like other settings. Once CalMAN has been restarted it always seems to default to being on.

For greyscale, I then found that dE u'v' worked best. I don't have any control over gamma so it's not relevant to greyscale performance in my case—I only want to see the colour error, and dE u'v' is the only one that takes luminance out of the equation with gamma correction disabled.

The problem with using dE u'v' is at low light levels it is over reporting what the error really is because it's an absolute error on chroma dC. The lower a light level is the harder it is for us to see a chroma error because of the way the eye switches from photopic to mesopic and finally scotopic. So as light levels approach black the chroma error dC does not matter as much because we can’t see it and at some point especially for red we don’t see it at all. With dE76 the chroma dC is used weighted to the light level error dL so that it estimates more how our eyes work. With dE94 it adds the ability to split the chroma dC into saturation dC and hue dH and the dL weighting is a bit different. Our bullseye charts are based on dE ‘u’v because you are only looking for the chroma error, the inner ring is dE 3.

These are all just tools to tell us when enough is enough and we are done because it won’t matter because we can’t see it.

Yes and that is what we recommend for use within CalMAN.

I thought this setting was global for the session (or all sessions).. is this incorrect?

Since the raw data is saved, I assumed when you changed that option both the grayscale and color would change to use that formula.. I also thought the gamma correction option was handled the same way (global for the session).

I thought this setting was global for the session (or all sessions).. is this incorrect?

Since the raw data is saved, I assumed when you changed that option both the grayscale and color would change to use that formula.. I also thought the gamma correction option was handled the same way (global for the session).

Yes both are global settings and need to be switched as needed.

You don't need CIE94 to get an analysis of LCH (and saturation) performance. LCH data can be easily derived from either CIELUV or CIELAB data, though CIELAB is not an appropriate metric for determining errors in saturation.

Because of this, the main advantage of CIE94 is not its inclusion of dL, dC, and dH, though it is handy.

CIE94 (and CIEDE2000) treat lightness errors very differently from either of the 1976 formulas. In a nutshell, the 1976 formulas predict that color error due to excessive saturation can be greatly reduced by lowering lightness below its nominal value. CIE94 predicts the opposite. The differences between them can in these cases be quite profound. There may be other differences as well, but this is the most significant one that I've run across.

It is also important to keep in mind that there are two sources of variance between the 1976 formula and subsequent versions. The 1976 color difference formula can use either L*a*b* or L*u*v* data. CIE94 and CIEDE2000 use L*a*b* exclusively.

The problem with using dE u'v' is at low light levels it is over reporting what the error really is because it's an absolute error on chroma dC. The lower a light level is the harder it is for us to see a chroma error because of the way the eye switches from photopic to mesopic and finally scotopic. So as light levels approach black the chroma error dC does not matter as much because we can’t see it and at some point especially for red we don’t see it at all. With dE76 the chroma dC is used weighted to the light level error dL so that it estimates more how our eyes work. With dE94 it adds the ability to split the chroma dC into saturation dC and hue dH and the dL weighting is a bit different. Our bullseye charts are based on dE ‘u’v because you are only looking for the chroma error, the inner ring is dE 3.

These are all just tools to tell us when enough is enough and we are done because it won’t matter because we can’t see it.

Thanks for the explanation Derek. I've certainly noticed that errors close to black (particularly too much red) do not matter as much as brighter ones, and that explains what I was seeing when comparing the different dE formulas.

That said, as my current displays all have fairly limited controls, I'm not too concerned about what the dE number specifically is, but rather, I try to get it as low as possible. E.g. I'm not concerned if I have a dE of 4, as long as that is the lowest I can get it using the coarse controls on a screen.

Would it be possible to have CalMAN ignore the gamma correction when dealing with colour in future versions? It only seems to apply to greyscale, and I find it's a pain having to toggle it on/off (and remembering whether it's on/off) which is why I have been using dE u'v' recently. (I use a layout that has everything I need on one screen rather than separate ones for colour/greyscale)

There is a problem for those that have everything on one report it sounds like...

Also, when you regenerate the session report (I also have a report with both grayscale and color in one) and disable gamma correction, won't it affect both grayscale and color in the current version (this applies to the current session as well)?

The problem with using dE u'v' is at low light levels it is over reporting what the error really is because it's an absolute error on chroma dC. The lower a light level is the harder it is for us to see a chroma error because of the way the eye switches from photopic to mesopic and finally scotopic. So as light levels approach black the chroma error dC does not matter as much because we can’t see it and at some point especially for red we don’t see it at all. With dE76 the chroma dC is used weighted to the light level error dL so that it estimates more how our eyes work. With dE94 it adds the ability to split the chroma dC into saturation dC and hue dH and the dL weighting is a bit different.Derek, I have read this several times now and I can't quite make out what you are trying to say.

First, you say there are problems with dE u'v' having to do with low light levels, but we stay in photopic vision all the way down to 1 cd/m2. Mesopic and scotopic vision are of dubious relevance to grayscale readings, which would almost always occur at 1 cd/m2 or higher.

Second, you appear to draw a distinction between dE u'v' and dE76, but what is the difference? Isn't dE u'v' just the CIELUV 1976 standard minus the lightness component, which is not relevant to grayscale readings in any case?

Third, as I point out below, you can easily breakout LCH errors from CIELUV or CIELAB. You don't need CIE94 for this.

Fourth, I really don't understand what you mean when you refer to "split[ing] the chroma dC into saturation dC and hue dH". Chroma is a mixture of lightness and saturation, not of hue and saturation.

Finally, since CIE94 is L*a*b*-based, it is a poor choice for deriving saturation in any case (CIE never recommended a chromaticity diagram for L*a*b*).

Could you please clarify?

Derek, I have read this several times now and I can't quite make out what you are trying to say.

First, you say there are problems with dE u'v' having to do with low light levels, but we stay in photopic vision all the way down to 1 cd/m2. Mesopic and scotopic vision are of dubious relevance to grayscale readings, which would almost always occur at 1 cd/m2 or higher.
Color errors are harder to see as light levels decrease, generally-speaking. A 4 dC* error in chrominance is going to be much easier to see at L* = 100 than at L* = 2. This is the point of the conversion from u'v' to u*v*. As you point out, this is before we bring the three modes of the HVS into the discussion.

Second, you appear to draw a distinction between dE u'v' and dE76, but what is the difference? Isn't dE u'v' just the CIELUV 1976 standard minus the lightness component, which is not relevant to grayscale readings in any case?
The dE(uv) method defines L* == 100. As a result, it systematically overstates total error (dE). As someone correctly points out, above, this makes it a the most strict method, especially as you get away from pure white (L* = 100). This is why we developed a way to eliminate the amount of color error attributable to gamma. It is much closer to a technically accurate measure of perceptible color error, even though the implementation itself has a bug.

Third, as I point out below, you can easily breakout LCH errors from CIELUV or CIELAB. You don't need CIE94 for this.
Yes, but not "for free". The methods and results are a bit different, and the absolute error numbers are a bit different, but the intuition given is largely the same. What you want at the end is a color error that is imperceptible for moving images with no referent.

Fourth, I really don't understand what you mean when you refer to "split[ing] the chroma dC into saturation dC and hue dH". Chroma is a mixture of lightness and saturation, not of hue and saturation.
This is where the terms "chroma" and "luma" really hinder the conversation. Technically, chrominance does not have any lightness in it. It gets confused with the color decoder concepts because of how a saturation control works in a color decoder (a scaler applied to both color difference channels). Saturation is the distance away from a target, and hue is the angle versus the ray against white. This is easier to represent using a Lab or Luv diagram, rather than a chromaticity diagram, but the concept is the same, even if the math is muddier.

Finally, since CIE94 is L*a*b*-based, it is a poor choice for deriving saturation in any case (CIE never recommended a chromaticity diagram for L*a*b*).
Technically, the CIE doesn't like chromaticity diagrams at all. The video engineers do like them, and they make nice charts for comparing relative gamut sizes -- assuming you use one that is vaguely perceptually uniform (i.e., not xy).

There is a problem for those that have everything on one report it sounds like...

Also, when you regenerate the session report (I also have a report with both grayscale and color in one) and disable gamma correction, won't it affect both grayscale and color in the current version (this applies to the current session as well)?

This is a long-standing bug in my algorithm that I need to take some time to fix in a way that won't make users scream. :( Until then, yes, turn off gamma correction when doing gamut work.

CIE94 (and CIEDE2000) treat lightness errors very differently from either of the 1976 formulas. In a nutshell, the 1976 formulas predict that color error due to excessive saturation can be greatly reduced by lowering lightness below its nominal value. CIE94 predicts the opposite. The differences between them can in these cases be quite profound. There may be other differences as well, but this is the most significant one that I've run across.
In the 1976 formula, dC* tends to scale down faster than dL* scales up. This is most likely the issue you are seeing.

Almost three years to the day since I wrote this:
http://www.avsforum.com/avs-vb/showthread.php?p=6637132#post6637132

Bill, it is late but I'll try to be lucid.

Of course, color perception is less acute at low light levels than at higher light levels, but I still don't see how this is relevant to measuring grayscale dE error UNLESS the formula you use compares the actual measured L against the ideal L for a specified level of stimulus. This approach would include any gamma errors in one's dE calculation. The thing is, why would anyone ever do this? Conflating gamma errors with grayscale errors is in my opinion a methodological error. I have always used ideal L values for both test and reference for grayscale (not for color), in which case the level of stimulus has no effect on the dE value.

Regarding chroma, I think that we understand this term differently. The color science resources I have consulted define chroma as colorfulness of a color relative to a white similarly illuminated. This (http://www.huevaluechroma.com/082.php) is a reasonably good online reference as is this wiki page (http://en.wikipedia.org/wiki/Saturation_%28color_theory%29). Since colorfulness increases as illumination increases (even as saturation stays the same), chroma is a mixture of saturation and brightness. BTW, it occurs to me that this is just another way of stating the point you made in your post that the eye sees more color at higher levels of illumination.

The best real-world example is the standard Color control found on all NTSC displays. The color control is not a saturation control. It is a chroma control. Lower it and the brightness of color AND its saturation goes down. On the other hand, lower saturation on a well-designed CMS, brightness does not go down. In fact, because you move the color towards the white point, it may go up somewhat.

Bill, it is late but I'll try to be lucid.
I'm often not coherent, myself, and I usually don't have a late hour to blame. :)

Of course, color perception is less acute at low light levels than at higher light levels, but I still don't see how this is relevant to measuring grayscale dE error UNLESS the formula you use compares the actual measured L against the ideal L for a specified level of stimulus.
This is the dL portion of dE. Also, you use a ratio of the actual Y vs. the ideal or reference Y to compute the L* scaler for converting u'v' into u* v*.
This approach would include any gamma errors in one's dE calculation. The thing is, why would anyone ever do this? Conflating gamma errors with grayscale errors is in my opinion a methodological error.
Agreed. The dE(u'v') is one attempt to get around this, but it has a few unintended side effects. Our gamma correction is another. Do be aware, though, that not all of dL (and dC, for that matter) is attributable to gamma error, though in practice, a lot is. Measurement error also factors in here, as well as. Depending upon whether you are talking about a primary or a white point, you also can end up with "color temperature" error, which will have both a chromaticity component AND a luminance component.
I have always used ideal L values for both test and reference for grayscale (not for color), in which case the level of stimulus has no effect on the dE value.
One thing to watch out for here is that you may have a gamma error that is isolated to a single channel, so factoring out all dL may hide some issues if you are really trying to dial a display in to the nth degree.

Regarding chroma, I think that we understand this term differently. The color science resources I have consulted define chroma as colorfulness of a color relative to a white similarly illuminated.
There is a movement afoot to try to drive some clarity in language when describing color in video engineering. Given the persistence of things like "IRE", it is a movement I support. In this case, chroma is to chrominance what luma is to luminance. In other words, common usage generally gives you enough context to determine whether you are talking about color as a physical property or color as part of a digital signal, but one really should not have to guess. In video engineering terms, chrominance is what we are really talking about.

This (http://www.huevaluechroma.com/082.php) is a reasonably good online reference as is this wiki page (http://en.wikipedia.org/wiki/Saturation_%28color_theory%29). Since colorfulness increases as illumination increases (even as saturation stays the same), chroma is a mixture of saturation and brightness. BTW, it occurs to me that this is just another way of stating the point you made in your post that the eye sees more color at higher levels of illumination.
Let's take a look at the introductory sentence in the wikipedia article:
In colorimetry (http://en.wikipedia.org/wiki/Colorimetry) and color theory (http://en.wikipedia.org/wiki/Color_theory), colorfulness, chroma, and saturation are related but distinct concepts referring to the perceived intensity of a specific color (http://en.wikipedia.org/wiki/Color). Colorfulness is the difference between a color against gray (http://en.wikipedia.org/wiki/Grey). Chroma is the difference of a color against the brightness (http://en.wikipedia.org/wiki/Brightness) of another color which appears white under similar viewing conditions. Saturation is the difference of a color against its own brightness.[1] (http://en.wikipedia.org/wiki/Saturation_%28color_theory%29#cite_note-0) Though this general concept is intuitive, terms such as chroma, saturation, purity, and intensity are often used without great precision, and even when well-defined depend greatly on the specific color model (http://en.wikipedia.org/wiki/Color_model) in use.

In the above quote, the only difference between the definitions of colorfulness and chroma is the white point to which the eye is adapted. Both are notionally independent of luminance, Grassman's law notwithstanding. The definition of saturation here depends heavily upon what the definition of "brightness" is. Poynton's definition, from the Color FAQ (which people can get easily) and referencing the CIE (whose documents are not necessarily easy to find) is a bit more clear:
6. What is saturation?
Again from the CIE, saturation is the colourfulness of an area judged in proportion to its brightness. Saturation runs from neutral gray through pastel to saturated colors. Roughly speaking, the more an SPD is concentrated at one wavelength, the more saturated will be the associated color. You can desaturate a color by adding light that contains power at all wavelengths.

Note the interplay of colorfulness here with saturation. In other words, the further you move away from white, the more saturated a color is. Looking further down the wikipedia article, though:

The naïve definition of saturation does not specify its response function. In the CIE XYZ and RGB color spaces, the saturation is defined in terms of additive color mixing, and has the property of being proportional to any scaling centered at white or the white point illuminant. However, both color spaces are nonlinear in terms of psychovisually perceived color differences (http://en.wikipedia.org/wiki/Color_difference). It is also possible, and sometimes desirable to define a saturation-like quantity that is linearized in term of the psychovisual perception.

The article then goes on to give the definition of the LCH coordinate system, where C* essentially corresponds to saturation. This is consistent with Poynton, Lindbloom, et al. The net effect is that there is some component of saturation that is dependent upon L*, it is desirable to minimize this effect as much as possible to get to a more internally consistent definition of Hue, Saturation and Lightness (HSL).

The best real-world example is the standard Color control found on all NTSC displays. The color control is not a saturation control. It is a chroma control. Lower it and the brightness of color AND its saturation goes down. On the other hand, lower saturation on a well-designed CMS, brightness does not go down. In fact, because you move the color towards the white point, it may go up somewhat.

This is where the use of the word chroma really hurts us. I would refer people who want a more detailed explanation to look at the help files in CalMAN under setting up a color decoder. I have tweaked these a bit for the upcoming v3.2 release to make some points a little more clear, but the core content remains the same. However, your understanding of what a color control does is incorrect.

In an analog environment, the color control operates on the amplitude of the color subcarrier. The tint control affects the phase angle of the color subcarrier. What this means for digital signals is that a saturation/color control ought to operate as a scaler to the color difference channels (back to that amplitude thing), raising or lowering them together. The tint or hue control changes the relative balance of the color difference channels (emphasizing or de-emphasizing one over the other). Say, for example, that you fed a pure red signal into a TV. Using real numbers to simplify the math, you get:

RGB(1.0, 0.0, 0.0) --> YCbCr709(0.2126, -0.1146, 0.5)

Now, let's turn the (ideal) saturation control down by half (multiply by 0.5). This is what you would get:

YCbCr709(0.2126, -0.0573, 0.25) --> RGB(0.6063, 0.1063, 0.1063)

Note what has happened in the Green and Blue channels. We have de-saturated red (moved it closer to white), but overall light levels stay the same (luma is constant, after all). How to reconcile the above with what you are seeing in practice? The answer has to do with the locations of the primaries, themselves. If your primaries are not perfect, then you may see a change in luminance when you change the relative mix of RGB (back to Grassman's Law).

One note: the above holds only for well-designed color decoders (one could conceivably use an adder, rather than a scaler, but I think scalers are far, far more common). For poorly-designed color decoders, anything goes.

I'll leave CMSs alone for now, but they also beg some clarification.

Bill

I just took a quick scan through the last page of Tom's CMS thread, and I thought that I should point out that his advice here:

For grayscale dE calculations, the luminance values don't count, so just keep the Y value at 1.0. For calculating dE for the primary/secondary colors, you divide the color's Y by the Y of white at the same level of stimulus. The values should always be less than 1.0.

...produces an identical result as dE(u'v') for grayscale.

Bill, for the life of me I still don't understand the position you advocate regarding gray scale measurements and the level of light output.

Let's be real specific. Using CIELUV, the dE of x0.314, y0.351 in a SMPTE-C or Rec. 709 color space is 17.5. It is 17.5 whether measured at 20% stimulus or 100% stimulus, because the level of stimulus is not part of grayscale dE equations.

You SEEM to be saying that at low light levels you would get something different than what you would get at high light levels because of the eye's varying sensitivity to color at varying levels of illumination. The only way that fact would be relevant to grayscale dE calculations is if you are somehow incorporating the L component into the calculation, which is what I described as a methodological error. On the other hand, if you don't incorporate the L component, then the level of stimulus simply wouldn't matter (1.0 - 1.0 is always zero). It would be 17.5 at all levels.

Regarding chroma, there are 2 issues: one theoretical and the other empirical. The theoretical part is a little hazy because even the passage I quoted concedes that the definition of chroma varies depending on the context. But the definition I have been using is colorfulness of a color relative to a sample of white at a similar level of illumination. Using this df, if you increase the brightness of a red alongside a patch of white whose brightness is similarly increased, the red will appear more colorful. If you simultaneously dim the red and white patches, the red would appear less colorful. Thus, chroma and brightness go together. However, there is enough confusion about the df of this term that I wouldn't press the theoretical point with much enthusiasm.

However, what I am more confident of is the empirical side. Color controls always adjust saturation AND brightness. I have never come across a Color control that operated purely in the saturation realm. In this sense it is a chroma control using the df I use above. In fact, small adjustments to the Color control adjust color brightness almost exclusively.

If you turn the Color control down to zero RGBCYM lose all of their colorfulness and much of their brightness. You are left with a series of grays of varying brightness. There must be some mathematical relationship between these residual grays, but I haven't been able to figure out what it is.

I just took a quick scan through the last page of Tom's CMS thread, and I thought that I should point out that his advice here:
...produces an identical result as dE(u'v') for grayscale.Yes. I guess I don't see your point. I wasn't arguing against using what you you refer to as dE(u'v'). What I was arguing against was the suggestion that we should somehow adjust our dE calculations to accommodate varying levels of stimulus.

Bill, for the life of me I still don't understand the position you advocate regarding gray scale measurements and the level of light output.
Okay. As I said, I can often be obtuse without trying.

Let's be real specific. Using CIELUV, the dE of x0.314, y0.351 in a SMPTE-C or Rec. 709 color space is 17.5. It is 17.5 whether measured at 20% stimulus or 100% stimulus, because the level of stimulus is not part of grayscale dE equations.This is not correct, strictly speaking. To compute dE, you would need to specify a measured Y (normalized) and an assumed gamma to compute dE in the above situation. Again, strictly speaking, color error is comprised of two parts: dL and dC in CIELuv. dL is the error in Lightness, or quantity of light, and dC is the error in chrominance (the uv portion).

Bruce Lindbloom supplies the math for computing dE using Lab, which is nearly identical (it is a Cartesian distance). You just need to replace the a* and b* parameters with u* and v*:
http://www.brucelindbloom.com/index.html?Eqn_DeltaE_CIE76.html
Computing Luv is pretty easy as well:
http://www.brucelindbloom.com/Eqn_XYZ_to_Luv.html

If you want to double-check the conversion or the math, take a look at sections 4 and 35 in Poynton's (http://www.poynton.com) Color FAQ. If you want to fact-check further what I said about Lindbloom, above, with Poynton, then grab a copy of Poynton's book, page 227, equation 21-9.

So, my question back to you is what are these special grayscale dE equations that are accepted by the CIE that eliminate L*? I think we all agree that isolating gamma error from "true" color error is a desirable thing, but your challenge on the technical front seems ill-informed or inarticulate. There are methods for isolating gamma error, but these are not defined standards because gamma error is part of color error. If you want to discuss the relative merits of particular methods of eliminating gamma error, then by all means, let's do so.

You SEEM to be saying that at low light levels you would get something different than what you would get at high light levels because of the eye's varying sensitivity to color at varying levels of illumination. The only way that fact would be relevant to grayscale dE calculations is if you are somehow incorporating the L component into the calculation, which is what I described as a methodological error. On the other hand, if you don't incorporate the L component, then the level of stimulus simply wouldn't matter (1.0 - 1.0 is always zero). It would be 17.5 at all levels.I won't address the preamble, since I've covered it (we are just dealing with basic arithmetic). So that we are clear, we do need to specify what formula we are talking about. In the above, do you convert u' and v' into u* and v* by multiplying by 13L*? Or are you taking a straight Cartesian distance in the u'v' plane? If the latter, you do not get to 17.5. If the former, well, it's easily demonstrable. Let's assume a constant 0.01 du' and dv', but let's vary L* (dL = 0):
L* dE
100 18.4
90 16.5
80 14.7
70 12.9
60 11.0
50 9.2
40 7.4
30 5.5
20 3.7
10 1.8

No advocacy here, merely arithmetic. Whether it has to do with the transition between Photopic, Mesopic or Scotopic vision is irrelevant at a practical level. What matters is the equation that is defined by the CIE. The intuition is easy: the more light you have, the easier it is to see a given difference in chrominance. How well this tracks reality is a subject for a different debate.

I, gently, remade this point a year and a half ago in your CMS thread:
http://www.avsforum.com/avs-vb/showthread.php?p=10652296#post10652296

Regarding chroma, there are 2 issues: one theoretical and the other empirical. The theoretical part is a little hazy because even the passage I quoted concedes that the definition of chroma varies depending on the context. But the definition I have been using is colorfulness of a color relative to a sample of white at a similar level of illumination. Using this df, if you increase the brightness of a red alongside a patch of white whose brightness is similarly increased, the red will appear more colorful. If you simultaneously dim the red and white patches, the red would appear less colorful. Thus, chroma and brightness go together. However, there is enough confusion about the df of this term that I wouldn't press the theoretical point with much enthusiasm.Then let's move on.

However, what I am more confident of is the empirical side. Color controls always adjust saturation AND brightness. I have never come across a Color control that operated purely in the saturation realm. In this sense it is a chroma control using the df I use above. In fact, small adjustments to the Color control adjust color brightness almost exclusively.Sigh. Having said that you are not going to press the point, you are now, well, pressing the point. This means that we need to get clear on definitions. Let's start here:

1) The use of a Color or Saturation control in an analog NTSC system is to control the amplitude or excursion of the color subcarrier (or color "sub-channel)?

2) The use of a Tint or Hue control in an analog NTSC system is to control the phase angle of the color subcarrier?

3) The Color and Tint are "color decoder" controls in that they control how the component video signal is decoded into RGB?

If we can't agree on the above, then we aren't going to get anywhere at all, let alone anywhere quickly.

If you turn the Color control down to zero RGBCYM lose all of their colorfulness and much of their brightness. You are left with a series of grays of varying brightness. There must be some mathematical relationship between these residual grays, but I haven't been able to figure out what it is.At it's most foundational, it's called "Grassman's Law". Then you have whatever computational quirks the engineers baked into the design when going from real numbers to integer math. On top of that, you have a liberal sprinkling of measurement error.

Bill

Yes. I guess I don't see your point. I wasn't arguing against using what you you refer to as dE(u'v'). What I was arguing against was the suggestion that we should somehow adjust our dE calculations to accommodate varying levels of stimulus.
Tom - If you are comfortable with the idea that your method overstates dE throughout the grayscale, then be fine with it. It creates a tough target, and one can have confidence that one is really below the threshold of visibility if you hit it. CalMAN allows people to use that method, but it can be a high hurdle for non-professional calibrators. We offer a different method for eliminating gamma error that does not have the same overestimation problem, with the desirable secondary effect of creating a more forgiving hurdle to assess whether you are over the threshold of visibility or not.

Let's be clear, though. You seem to be fighting a strawman that you have labelled with our names. If it weren't for the fact that other people are reading this (maybe none are, anymore), I wouldn't bother. You haven't addressed the points I have made above, but merely reasserted your own incomplete understanding. Sorry, but unless others want the discussion to continue, I feel like honor has been served for defending our methods.

I haven't responded because I have been on Thanksgiving vacation and not following the thread.

I really don't understand why these debates cannot remain respectful and without personal animus. It is not personal. It was not my intention to question your "honor" and I'll ignore your unhelpful assertions about what I do and do not understand.

Of course, the CIELUV and CIELAB specifications include consideration of the brightness of the measured sample compared to a reference. This is because the colors in a gamut include a specified level of brightness. Thus, a measured color whose brightness deviates from that reference has a higher dE than one that doesn't. However, the same cannot be said for white.

The basis of this debate is stated in your claim that "To compute dE, you would need to specify a measured Y (normalized) and an assumed gamma to compute dE in the above situation."

That is the entirety of this argument: should an assumed gamma and a measured Y be included in dE calculations for grayscale. Note: we need an assumed gamma because this the ONLY sense in which we could assume a "correct" level of brightness for white. Unlike color, the definition of white includes no specific level of brightness. Indeed, the specified level of brightness for color is defined relative to the brightness of reference white, which is NOT defined in advance.

So this all boils down to whatever reasons one can offer for or against including a measured Y and an assumed gamma in one's dE calculations for grayscale. The reasons against, as I have suggested, are compelling. Let me enumerate them:

1. Unlike color, there simply is no specification for the "correct" level of brightness for white. There is only what the target gamma specifies.

2. There is not even any fixed definition of the "correct" gamma. This is both because there is more than one way to calculate gamma and because gamma requires assumptions about viewing environment and the characteristics of the display on which the content is mastered.

3. The methodology you suggest is a literal application of the CIE specification for a context in which it was never intended. The CIE specification is intended for grading color difference, not for grayscale tracking, which poses a different set of issues.

4. The methodology you suggest is inconsistent with industry standards. There is no other source I am aware of that calculates grayscale error in the way you describe. ColorFacts does not. Greg Rogers' Display Calibration Calculator does not. ISF recommends that it is MORE important to get the low end of the grayscale right rather than LESS important as you seem to suggest. Cliff Plavin of Progressive Labs has always argued for dC only for grayscale. I have resisted this, because although I agree that dL should be ignored, I didn’t see any reason to ignore dH. However, when you look at the actual data, the contribution of dH to grayscale errors is negligible. Thus, I now think this is a defensible position. The one you advocate here is not. If you are aware of any other reputable source that calculates grayscale dE the way you advocate here, then I'd sure like to know.

5. This methodology has the effect of systematically and sometimes grossly understating grayscale error. One might argue the point because of the eye’s poor sensitivity to color at very low light conditions, as was suggested in the original post's irrelevant discussion of scotpic, mesopic, photopic vision. But as I pointed out, the level of illumination required for this phenomenon to matter is much lower than any grayscale reading would ever include. Nonetheless, you apparently continue to endorse the view that we should calculate the dE of a specified white, say x0.314, y0.351, differently at 90% stim than at 100% stim where our variable sensitivity to color perception is surely irrelevant. Your own example has the dE of the SAME COLOR OF WHITE dropping nearly 2 points for no other reason than from going from 100% to 90% stim and reduced by more than half at 40% stim. I frankly find it difficult to even take this very seriously. It appears to arise from a rigid application of a mathematical model without regard to the consequences for practical application much less common sense.

Consider the following. Assume a 35 fL peak output, 2.2 gamma display. Also assume a consistent white of x0.314, y0.351. This is the white point for DCI. Now what’s the dE of this white relative to a Rec. 709 reference of x0.3127, y0.329? According to the methodology you endorse here, at 100% stim it is 17.5 (you get a slightly different number I suppose because of rounding differences), but at 20% stim it is 3.4, well within the SMPTE’s recommendation of a dE of no more than 4.0 L*a*b* units. So, using your recommended methodology, x0.314, y0.351 at 20% stim is a perfectly acceptable calibration result. Here’s the RGB values for that color of white.

http://home.comcast.net/%7Etlhuffman/rgb_grayscale.gif

Bill, this is madness. Are you really telling your customers to not bother with an error this large at the low end of the grayscale because our eyes just won’t be sensitive to it? Here’s a simple experiment. Just look at a 20% (approx. 3.5 cd/m2) window or field of x0.314, y0.351. If you can’t tell the difference between that and another window or field of x0.3127, y0.329 then getting your display calibrated is the least of your worries.

6. This recommendation isn't even internally inconsistent. Surely you do not also advocate using a different set of standards for judging the dE of pri/sec colors when using 75% windows as opposed to 100% patterns. But you should to be consistent with your view that the eye becomes progressively less sensitive to color at lower light levels and that the dE measurement should reflect this. If you did then, all else being equal, a color of red at 75% should have a lower dE than the same color of red at 100% stim. Of course, you don’t advocate this because the brightness of color is judged against white as a reference, which changes proportionally. But that’s the point, isn’t it? The brightness of white is relevant for gamma, levels, and calculating the dE of color. It is NOT relevant for calculating the dE of itself.

You seem to want to make the narrow point that IF you include a measured lightness element and assumed gamma in dE calculations that you get the result you describe. I don’t question this result. What I question is whether this approach offers a reasonable calibration methodology or sound color science.

BTW, Grassmann's Law explains why the brightness of RGB equals the brightness of white and why the brightness of each of the secondaries equals the brightness of the contributing primaries. That was not the issue I mentioned, which was the relative brightness (relative to each other and relative to the original white at the same level of stimulus) of the grays that result after removing all of the chroma from color signals.

That is the entirety of this argument: should an assumed gamma and a measured Y be included in dE calculations for grayscale.
Let's recap.
Yes both are global settings and need to be switched as needed.
Referring to gamma correction and dE formula in CalMAN. Given that we have gamma correction "on" by default might tell you something about our position on gamma error when measuring grayscale color error.
Second, you appear to draw a distinction between dE u'v' and dE76, but what is the difference? Isn't dE u'v' just the CIELUV 1976 standard minus the lightness component, which is not relevant to grayscale readings in any case?
At which point, Derek asked me to step in to discuss the technical details in the calculations. Since you asked about our methods, I thought I would talk about them. Since your operating definition of saturation does not jibe with how color scientists use it, I felt compelled to correct you on it, lest people start abusing "saturation" the way they do "IRE" (i.e., non-specific, and largely divorced from its original intent or context).
This is why we developed a way to eliminate the amount of color error attributable to gamma. It is much closer to a technically accurate measure of perceptible color error, even though the implementation itself has a bug.
Here's where your strawman evaporates. Yes, gamma error ought to be separable from the rest of color error. The issue is now one of method.
Conflating gamma errors with grayscale errors is in my opinion a methodological error.
We are agreed, but you go on.
I have always used ideal L values for both test and reference for grayscale (not for color), in which case the level of stimulus has no effect on the dE value.
This is not true. The level of stimulus you assume inflates dE the further away from white you get. It is a systematic bias upwards on the calculation. Then you venture off with more chroma/saturation silliness.
Agreed. The dE(u'v') is one attempt to get around this, but it has a few unintended side effects. Our gamma correction is another.
<sarcasm>Yep, we are definitely at loggerheads here over whether to include gamma error in grayscale dE calculations.</sarcasm> :rolleyes:
Bill, for the life of me I still don't understand the position you advocate regarding gray scale measurements and the level of light output.
I hope it is now PAINFULLY clear to all, including you.
You SEEM to be saying ...
This is the strawman.
Yes. I guess I don't see your point.
Correct, and you obviously still don't.

I think we all agree that isolating gamma error from "true" color error is a desirable thing, but your challenge on the technical front seems ill-informed or inarticulate. There are methods for isolating gamma error, but these are not defined standards because gamma error is part of color error. If you want to discuss the relative merits of particular methods of eliminating gamma error, then by all means, let's do so.
It's like I'm talking to a wall.

So, having cleared up the great "gamma/grayscale color error debate of 2008", let's have some fun, shall we?

Bill

Since your operating definition of saturation does not jibe with how color scientists use it, I felt compelled to correct you on it, lest people start abusing "saturation" the way they do "IRE" (i.e., non-specific, and largely divorced from its original intent or context).Bill, you are just making this stuff up. Please refer us to where I defined "saturation", incorrectly or otherwise. What we mildly disagreed about was the definition of chroma, which you are apparently confusing with saturation. I defined chroma as "colorfulness of a color relative to a sample of white at a similar level of illumination." Here's how Fairchild defines chroma "Colorfulness of an area judged as a proportion of the brightness of a similarly illuminated area that appears white. . ." [Color Appearance Models, p. 103]. Here's how Schanda defines chroma: "Colorfulness of an area judged as a proportion of the brightness of a similarly illuminated area that appears white. . ." [Colormetry: Understanding the CIE System, Glossary, p. 445]. The two online references I provided offered similar definitions. I could go on. However, if I were to define saturation it would be either "colorfulness relative to its own brightness", or, more simply, "distance from the white point on a CIE chart."

Here's where your strawman evaporates. Yes, gamma error ought to be separable from the rest of color error. The issue is now one of method.A strawman is when one incorrectly characterizes his opponent's position or argument so as to make it seem less plausible than it really is. The only position that I attribute to you for the purpose of this argument is the one I quoted directly, namely that "To compute dE, you would need to specify a measured Y (normalized) and an assumed gamma to compute dE in the above situation."

This is not true. The level of stimulus you assume inflates dE the further away from white you get. It is a systematic bias upwards on the calculation.I feel a need to remind everyone that assertion is not the same as argument. I gave a series of reasons for why one should not use an assumed gamma and measured brightness in dE calculations for grayscale in my latest post, including examples of others who I guess you think are similarly mistaken. Instead of responding, you simply assume the truth of the very position in dispute and offer a series of truncated quotes from earlier posts, which has the overall effect of making it rather difficult to make sense of this latest post.

Then you venture off with more chroma/saturation silliness.It seems silly to you because you apparently don't understand the difference between chroma and saturation. I refer you to the sources I have already cited--online and published hardcopy--for clarification. I will gladly provide others.

I am often amused by the extremely technical and razor thin distinctions that we often obsess over in this forum. Debates about whether we should use 2.5 or 2.2 as a target gamma is one example. Kraz and I have debated intensely about whether to use CIELUV or CIELAB. The thing is, these are relatively small distinctions. What is breathtaking about this debate is that the methodology that you endorse renders truly enormous grayscale errors perfectly acceptable.

I asked before, and you chose not to respond. Can you cite any other reputable source that would grade a 20% stim (3.5 cd/m2) grayscale reading of x0.314, y0.351 as within acceptable dE tolerances for a Rec. 709 reference? Since you obviously have no respect for my view on the subject and don't seem particularly interested in the sources I cite, then perhaps Greg Rogers' Display Calibration Calculator (http://www.accupel.com) might interest you.

http://home.comcast.net/%7Etlhuffman/rogers1.gif

You would suggest that it would be perfectly valid to grade the grayscale dE error at 20% stim as 3.4, instead of 17.5. This is not a minor technical discrepancy. This is night and day.

BTW, one correction. It turns out that Grassmann's Law can explain the phenomenon I was describing. The luminance of the 3 primaries turned into gray will equal 255 in RGB. The proportions are R0.299, G0.587, B0.114, which equal 1.0. However, I still don't see it in CIE Y.

I haven't responded because I have been on Thanksgiving vacation and not following the thread.

I really don't understand why these debates cannot remain respectful and without personal animus. It is not personal. It was not my intention to question your "honor" and I'll ignore your unhelpful assertions about what I do and do not understand.
Okay, so how doesn't your method of eliminating color error overestimate dE at, say, 30% for a given absolute difference in u' and v'?

Of course, the CIELUV and CIELAB specifications include consideration of the brightness of the measured sample compared to a reference. This is because the colors in a gamut include a specified level of brightness. Thus, a measured color whose brightness deviates from that reference has a higher dE than one that doesn't. However, the same cannot be said for white.White is, um, definitional. It's a normalization thing. Yn == 1.0.

The basis of this debate is stated in your claim that "To compute dE, you would need to specify a measured Y (normalized) and an assumed gamma to compute dE in the above situation."

That is the entirety of this argument: should an assumed gamma and a measured Y be included in dE calculations for grayscale.Good. This is where it gets fun, because the "real" "debate" seemed to be whether eliminating gamma error was a good thing, and since that debate is settled, let's talk specific implementations. :)
Note: we need an assumed gamma because this the ONLY sense in which we could assume a "correct" level of brightness for white.White is white. It's a definitional thing. ALL luminance values need to be normalized against the luminance for white, which is 1.0 (what "normalization" means).
Unlike color, the definition of white includes no specific level of brightness. Indeed, the specified level of brightness for color is defined relative to the brightness of reference white, which is NOT defined in advance.Wow. Simply wow. Which definition of color includes a specific level of brightness? Is red like 3.14159 ftL? Does it turn magenta when it gets to 9 ftL? Does color care whether you are in metric or imperial measures?

Here's a hint: all luminance values need to be normalized against white. That is the definition. As a result of this, gamma is undefined at Yn = 1.0.

Now gray, on the other hand does depend upon what flavor of gray you are trying to produce. In other words, you need to know some things about what was intended versus what was actually produced.

So this all boils down to whatever reasons one can offer for or against including a measured Y and an assumed gamma in one's dE calculations for grayscale. The reasons against, as I have suggested, are compelling. Let me enumerate them:

1. Unlike color, there simply is no specification for the "correct" level of brightness for white. There is only what the target gamma specifies.I'm done chiding you for being sloppy on distinguishing gray from white. The "correct" level of gray is a combination of the target gamma formula, the target exponent and the signal level.

2. There is not even any fixed definition of the "correct" gamma. This is both because there is more than one way to calculate gamma and because gamma requires assumptions about viewing environment and the characteristics of the display on which the content is mastered. Good reasons to eliminate gamma error, but then computing gamma error, if it is so unknown, should be pretty tough, huh?

3. The methodology you suggestYou have clearly demonstrated that you do not know this.
is a literal application of the CIE specification for a context in which it was never intended.Huh? They did not know about gamma and grayscale in the 70s?
The CIE specification is intended for grading color difference,Of which, grayscale is an example.
not for grayscale tracking, which poses a different set of issues. Such as?? I'm really curious. This is stuff that seems not to have been included in the SMPTE Journal.

4. The methodology you suggest is inconsistent with industry standards. There is no other source I am aware of that calculates grayscale error in the way you describe.Aside from the CIE, right? But hey, what do they know? :) In other words, show me how you aren't overestimating error? Or concede that it is something you aren't worried about.

By the way, which method are we talking about, again? I'm not sure since I'm pretty sure we have not disclosed our actual algorithm, pretty much ever. And we most assuredly have not done so in this thread.
ColorFacts does not.We take pride in having deployed things like point gamma calculations years before they did. This really isn't any different.
Greg Rogers' Display Calibration Calculator does not.Greg generally uses dE(uv).
ISF recommends that it is MORE important to get the low end of the grayscale right rather than LESS important as you seem to suggest.You are making this too easy:
What's wrong with the ISF description of color?

I am a graduate of the ISF seminar, and I think that the organization has performed a valuable service at educating the public about the importance of accurate video. However, the ISF understanding of color is not entirely clear.

Cliff Plavin of Progressive Labs has always argued for dC only for grayscale. I have resisted this, because although I agree that dL should be ignored, I didn’t see any reason to ignore dH. However, when you look at the actual data, the contribution of dH to grayscale errors is negligible. Thus, I now think this is a defensible position.Really? You think that dE76 doesn't encompass dH inside dC? Would it surprise you to learn that it does?
The one you advocate here is not. If you are aware of any other reputable source that calculates grayscale dE the way you advocate here, then I'd sure like to know.This is getting pedantic, but the only methods I have talked about are either a) the standards themselves, or b) people's specific "adjustments" to them.

5. This methodology has the effect of systematically and sometimes grossly understating grayscale error. One might argue the point because of the eye’s poor sensitivity to color at very low light conditions, as was suggested in the original post's irrelevant discussion of scotpic, mesopic, photopic vision. But as I pointed out, the level of illumination required for this phenomenon to matter is much lower than any grayscale reading would ever include. Nonetheless, you apparently continue to endorse the view that we should calculate the dE of a specified white, say x0.314, y0.351, differently at 90% stim than at 100% stim where our variable sensitivity to color perception is surely irrelevant.Tom - Please stop. Seriously. Reading comprehension is not your strong suit. I'm having fun picking this apart, but now you are embarrassing yourself.

So that we are clear: the method should be constant across any set of numbers. The output is what changes based upon the inputs. All that I posted, above, was a set of numbers that were straight CIE76 with dL = 0 and a constant du' and dv'. That's all. The point was to demonstrate a desirable property of color error that was intended by the CIE, and that also shows up in CIE94 and CIE00. It is this property that our little trick preserves, while eliminating gamma error. No, we do not aribtrarily set dL to 0.

The rest of this is simply demonstrating that despite having published a dE calculator of your own, you do not really understand what goes on in the numbers. Your complaint, in this case, is with the CIE itself, not us..

Your own example has the dE of the SAME COLOR OF WHITE dropping nearly 2 points for no other reason than from going from 100% to 90% stim and reduced by more than half at 40% stim.Yep. The numbers are what they are, it's not our method. In fact, it is a lot closer to yours. The only thing I did was let L* vary, rather than hold it constant to show how the CIE intended the equation to work. Obviously, you did not recognize that.

If you think it is as easy to see a given color difference when there is approx. 1/7th - 1/10th as much light around as what you are adapted to for white then that sounds like a testable hypothesis. One might even develop a model, of people's color perception. Here's a hint: if you want to be taken seriously, don't use your "sunlight/inside light" example from your WSR article. That one was simply bad.

I frankly find it difficult to even take this very seriously. It appears to arise from a rigid application of a mathematical model without regard to the consequences for practical application much less common sense. What consequences are there for practical application? That you set a high hurdle for imperceptibility? That you have to rely on software that does all of the computation for you in real time? That our users have to toggle a control if they want to use it? Do tell. I'm curious here, too.

As for common sense: I'm all ears. I'd really be entertained to learn why you prefer tool vendors who were fast-and-loose with published standards.

Consider the following. Assume a 35 fL peak output, 2.2 gamma display. Also assume a consistent white of x0.314, y0.351. This is the white point for DCI. Now what’s the dE of this white relative to a Rec. 709 reference of x0.3127, y0.329? According to the methodology you endorse here, at 100% stim it is 17.5 (you get a slightly different number I suppose because of rounding differences), but at 20% stim it is 3.4, well within the SMPTE’s recommendation of a dE of no more than 4.0 L*a*b* units. So, using your recommended methodology, x0.314, y0.351 at 20% stim is a perfectly acceptable calibration result.Careful, you are perilously close to saying that SMPTE uses L* when computing dE (you would be right there, though individuals vary in interpretation as this thread demonstrates). However, your math is off. Let's leave aside Lab76 alone until we master Luv. In CIE76 or CIE94 terms, it is over 9.

Here’s the RGB values for that color of white.Our numbers match here.

Bill, this is madness.Agreed.
Are you really telling your customers to not bother with an error this large at the low end of the grayscale because our eyes just won’t be sensitive to it?No.
Here’s a simple experiment. Just look at a 20% (approx. 3.5 cd/m2) window or field of x0.314, y0.351. If you can’t tell the difference between that and another window or field of x0.3127, y0.329 then getting your display calibrated is the least of your worries.Agreed. However, your "simple experiments", judging by the WSR article, often involve such "minor nuisances" as significant changes in ambient light spectrum and luminous flux, so one does have to double-check these things.

6. This recommendation isn't even internally inconsistent. Surely you do not also advocate using a different set of standards for judging the dE of pri/sec colors when using 75% windows as opposed to 100% patterns. But you should to be consistent with your view that the eye becomes progressively less sensitive to color at lower light levels and that the dE measurement should reflect this.Huh? It does, and so does yours, unless you use dE(uv) for gamut work. We recommend people turn off gamma correction when doing primary/secondary work. This was discussed previously.

If you did then, all else being equal, a color of red at 75% should have a lower dE than the same color of red at 100% stim.A given absolute difference in chromaticity would be less visible at lower light levels. Whether it drops below the threshold of visibility is another matter. This is curious from someone who just published something on the underappreciated nature of brightness in color measurement. Ironic, even.

Of course, you don’t advocate this because the brightness of color is judged against white as a reference, which changes proportionally. But that’s the point, isn’t it? The brightness of white is relevant for gamma, levels, and calculating the dE of color. It is NOT relevant for calculating the dE of itself.Huh? Are you confusing white, which is a point, with gray, which is a line connecting white and black? If white changes, then the entire adaptation model changes (must be re-normalized). Luminous flux is captured in the "Lightness" component of color error. You really don't understand this, do you?

You seem to want to make the narrow point that IF you include a measured lightness element and assumed gamma in dE calculations that you get the result you describe. I don’t question this result. What I question is whether this approach offers a reasonable calibration methodology or sound color science.Gauntlet thrown; gauntlet accepted. I am now free to post my critique of your CMS document. Hilarity will ensue.

BTW, Grassmann's Law explains why the brightness of RGB equals the brightness of white and why the brightness of each of the secondaries equals the brightness of the contributing primaries. That was not the issue I mentioned, which was the relative brightness (relative to each other and relative to the original white at the same level of stimulus) of the grays that result after removing all of the chroma from color signals.Tom - Here's a clue, as in "let me spell it out for you":

The additive mixture of primaries is an algebraic property between the defined primary locations and the white point. If your white point is constant (say, D65), but your primaries deviate from the established standard, then a change in the relative mix of the primaries will change the amount of total light, aka Grassman's law. When you change a saturation control, you are changing the relative mix of these primaries (note my red desaturation example, above). In other words, whether the predicted "no change" in luminance actually occurs is entirely dependent upon the actual locations of the primaries.

This application of Grassman's law is, I suspect, more advanced than your understanding since your CMS document includes the contradictory admonishments to set red to as close to 21% as possible using a saturation control, while not using color decoder controls to change the gamut. However, we'll give you a head start to make wholesale changes. It's late, and I am now bored with this utterly and thoroughly.

Bill

Bill, you are just making this stuff up. Please refer us to where I defined "saturation", incorrectly or otherwise. What we mildly disagreed about was the definition of chroma, which you are apparently confusing with saturation. I defined chroma as "colorfulness of a color relative to a sample of white at a similar level of illumination." Here's how Fairchild defines chroma "Colorfulness of an area judged as a proportion of the brightness of a similarly illuminated area that appears white. . ." [Color Appearance Models, p. 103]. Here's how Schanda defines chroma: "Colorfulness of an area judged as a proportion of the brightness of a similarly illuminated area that appears white. . ." [Colormetry: Understanding the CIE System, Glossary, p. 445]. The two online references I provided offered similar definitions. I could go on. However, if I were to define saturation it would be either "colorfulness relative to its own brightness", or, more simply, "distance from the white point on a CIE chart."
Tom - Please read. Really. Please read. What I said, above, was, essentially, that chroma has different meanings depending upon context, and these contexts have important and very different distinctions. To a video engineer, chroma is the color difference signal which pairs with luma. Poynton, chapter 11; I especially like the margin note on page 94 ("The video literature often calls these quantities chrominance. That term has specific meaning in color science, so in video I prefer the term modulated chroma.") The point here is that if you use chroma to mean both, then you need to be really clear about what it is you are talking about.

Please show me, in my quotes, where I confused chroma and saturation, with chroma as you define it. I'm waiting. I found this particularly cute in your wikipedia article:

"With three attributes—colorfulness (or chroma or saturation), lightness (http://en.wikipedia.org/wiki/Lightness_%28color%29) (or brightness), and hue (http://en.wikipedia.org/wiki/Hue)—any color can be described." (emphasis mine)

A strawman is when one incorrectly characterizes his opponent's position or argument so as to make it seem less plausible than it really is. The only position that I attribute to you for the purpose of this argument is the one I quoted directly, namely that "To compute dE, you would need to specify a measured Y (normalized) and an assumed gamma to compute dE in the above situation."
To which situation, then do you think I was referring? And which attributes did you attach to it?

I feel a need to remind everyone that assertion is not the same as argument. I gave a series of reasons for why one should not use an assumed gamma and measured brightness in dE calculations for grayscale in my latest post, including examples of others who I guess you think are similarly mistaken. Instead of responding, you simply assume the truth of the very position in dispute and offer a series of truncated quotes from earlier posts, which has the overall effect of making it rather difficult to make sense of this latest post.
Irony, Tom, irony. That post was my frustration that you were not actually reading nor responding directly to what I wrote. The post afterwards, which was much longer, took longer to write.

It seems silly to you because you apparently don't understand the difference between chroma and saturation. I refer you to the sources I have already cited--online and published hardcopy--for clarification. I will gladly provide others.

You mean the variable one you used initially, or the ones that repeat the definition you called, what was it, hazy?
The theoretical part is a little hazy because even the passage I quoted concedes that the definition of chroma varies depending on the context.
I understood it. At one point, you even believed that I did:
BTW, it occurs to me that this is just another way of stating the point you made in your post that the eye sees more color at higher levels of illumination.
In fact, you were perilously close to saving yourself my preceding post about how the dE equations work if you could have connected your quote, immediately above, with this:
Of course, color perception is less acute at low light levels than at higher light levels, but I still don't see how this is relevant to measuring grayscale dE error UNLESS the formula you use compares the actual measured L against the ideal L for a specified level of stimulus.
This all might have been over.

I am often amused by the extremely technical and razor thin distinctions that we often obsess over in this forum. Debates about whether we should use 2.5 or 2.2 as a target gamma is one example. Kraz and I have debated intensely about whether to use CIELUV or CIELAB. The thing is, these are relatively small distinctions. What is breathtaking about this debate is that the methodology that you endorse renders truly enormous grayscale errors perfectly acceptable.
If you could calculate them correctly, I'd understand, but the sub-text of the argument is that, well, you don't.

I asked before, and you chose not to respond. Can you cite any other reputable source that would grade a 20% stim (3.5 cd/m2) grayscale reading of x0.314, y0.351 as within acceptable dE tolerances for a Rec. 709 reference?
Asked and answered. >9 is not acceptable.
Since you obviously have no respect for my view on the subject
Not any longer.
and don't seem particularly interested in the sources I cite, then perhaps Greg Rogers' Display Calibration Calculator (http://www.accupel.com) might interest you.
He uses dE(uv). We covered this.

You would suggest that it would be perfectly valid to grade the grayscale dE error at 20% stim as 3.4, instead of 17.5. This is not a minor technical discrepancy. This is night and day.
Agreed, but this has been a tutorial, of sorts, on calculating dE. I hope you learn the lesson.

BTW, one correction. It turns out that Grassmann's Law can explain the phenomenon I was describing.
Hallelujah!
The luminance of the 3 primaries turned into gray will equal 255 in RGB.
So close. Yet so far.
The proportions are R0.299, G0.587, B0.114, which equal 1.0. However, I still don't see it in CIE Y.
Use the custom white point function in ColorFacts, since I assume this has put you off of our stuff, to compute what your color mix is for a given set of real primaries. Then draw a big circle around those values to account for 4%+ measurement error in luminance. Try a little harder; you may get there.

Okay, so how doesn't your method of eliminating color error overestimate dE at, say, 30% for a given absolute difference in u' and v'?Please stop referring to this as "my" method. This the method that the entire industry, so far as I can tell, except for you, uses for gray scale grading.

Good. This is where it gets fun, because the "real" "debate" seemed to be whether eliminating gamma error was a good thing, and since that debate is settled, let's talk specific implementations.This leaves me speechless. Bill, I cannot imagine where you would have gotten the idea that this debate was ever about whether "eliminating gamma error was a good thing."

Wow. Simply wow. Which definition of color includes a specific level of brightness? Is red like 3.14159 ftL? Does it turn magenta when it gets to 9 ftL? Does color care whether you are in metric or imperial measures?This misinterpretation of what I wrote is so bizarre that I can only assume that it is intentional. Yes, the definition of red includes a specific level of brightness. What is that level? Well, for Rec. 709 it is 21.26% of reference white. Now is it remotely possible that you didn't understand that this is what I was referring to, or were you intentionally mischaracterzing my well-known and often-repeated view just to try to score a point? What was it you referred to earlier? Straw man? This is a textbook example.

You have clearly demonstrated that you do not know this.
Huh? They did not know about gamma and grayscale in the 70s?
Of which, grayscale is an example.
Such as?? I'm really curious. This is stuff that seems not to have been included in the SMPTE Journal.Your unfortunate habit of responding individually to snippets of text rather than the entire pasage (or even complete sentence) makes responding to this difficult. Basically, I would just repeat what I said above. CIE has never, to my knowledge, published standards for how to implement dE specifically in grayscale tracking. That's not what they do. The implementation of CIE standards is up to individual industries.

By the way, which method are we talking about, again? I'm not sure since I'm pretty sure we have not disclosed our actual algorithm, pretty much ever. And we most assuredly have not done so in this thread.Bill, you really can't have it both ways. You cannot one moment refer to CIE standards and then in the next breath refer to "our actual algorithm" as though you have implemented something proprietary with respect to dE and grayscale tracking. The CIE dE formula is widely published and relatively simple. Our debate is about the implementation of that formula in grayscale tracking. Presumably we use the same forumla, except that your implementation of it--at least one version--includes the actual measured Y values which are compared against what the target gamma says they should be at each level of stimulus.

Greg generally uses dE(uv).Not just Greg. This or dE(ab) is what the enture industry uses for grayscale dE. That's what we are arguing about. I don't use the label dE(uv) or dE(ab) because I think people find it confusing. BTW, CIE never, so far as I know, endorsed anything called dE(uv), which is a perfect example of how individual industries adapt CIE standards to suit their own needs.

You are making this too easy:Are doing this on purpose? I can't quite tell if you are having us on at this point. In an attempt to show that I have contridicted myself, you quote me citing ISF one place and criticizing them another place. Unfortunately, for anyone who cares to read the full passage, which you (again) quote only the first couple of sentences of from an entirely different thread without any indication where this comes from, they will see that my criticism of ISF has to do with their vague distinction between color accuracy and color saturation. I cite them approvingly here for their training about the importance of getting the low end of the grayscale right. Is it possible for ISF to be mistaken in one area and correct in another? I think so.

Really? You think that dE76 doesn't encompass dH inside dC? Would it surprise you to learn that it does?No, Bill, it does not surprise me. I guess I don't see your point.

This is getting pedantic, but the only methods I have talked about are either a) the standards themselves, or b) people's specific "adjustments" to them.I asked a specific question, a question that you refused to acknowledge, must less answer. I'll ask it again. Do you know of any reputable source that uses dE the way you advocate here for grayscale measurements. Your silence indicates that no you do not.

Tom - Please stop. Seriously. Reading comprehension is not your strong suit. I'm having fun picking this apart, but now you are embarrassing yourself.

So that we are clear: the method should be constant across any set of numbers. The output is what changes based upon the inputs. All that I posted, above, was a set of numbers that were straight CIE76 with dL = 0 and a constant du' and dv'. That's all. The point was to demonstrate a desirable property of color error that was intended by the CIE, and that also shows up in CIE94 and CIE00. It is this property that our little trick preserves, while eliminating gamma error. No, we do not aribtrarily set dL to 0.The obnoxious hand waving aside, this is an important point that I want to emphasize. The quote you (thankfully) reproduce in full made 2 points. First, I mentioned the fact that Derek offered an irrelevant and misleading point about color perception at very low light levels as a way of justifying this methodology (a tactic you have had the good sense to not repeat). I only mentioned it because using this type of technical nonsense as a marketing tactic irritates me. Second, I wrote that "Nonetheless, you apparently continue to endorse the view that we should calculate the dE of a specified white, say x0.314, y0.351, differently at 90% stim than at 100% stim where our variable sensitivity to color perception is surely irrelevant." This was not phrased correctly. What I should have said was "Nonetheless, you apparently continue to endorse the view that we should calculate the dE of a specified white, say x0.314, y0.351, in such a way that at 90% stim you get different results than at 100% stim where our variable sensitivity to color perception is surely irrelevant."

Now here's a question for you and anyone else reading this. Given everything that has been written so far, what was more likely: a) that I was simply careless and misspoke; or b) that I truly didn't understand what you were doing?

Your natural inclination, of course, is to immediately respond based on b) because rather than argue the merits of this methodology, you insist upon simply attacking me. My understanding of CIE, my understanding of SMPTE, my understanding of your methods, etc. What's that old lawyer's adage? When the law is against you, argue the facts. When the facts are against you, argue the law. And when both are against you, attack your opponent.

There is nothing about this I don't understand. It is a simple point. The fact that you are not very enthusiastic about defending the methodology on its merits and resort instead to continually insisting that I lack understanding on this or that indicates the weakness of your position.

Here's CalMan's methodology (at least one of them) for calculating grayscale dE in the u'v' space. Assume a target gamma of 2.2, Rec. 709, and the following grayscale readings:
90% stim
x0.314
y0.351
Y0.793

20% stim
x0.314
y0.351
Y0.029

What's the dE? Well, using the method you endorse, the dE is 12.2 at 90% and at 4.1 at 20%. But, x0.314, y0.351 translates to
R 89%
G 105%
B 85%
and is an error large enough that it would be clearly visible at 20% stim and completely unacceptable as a calibration target. Yet, your dE metric says that the error is only 4.1, which is a (barely) acceptable result.

The formula is:
=SQRT((L1-L2)^2+(u1-u2)^2+(v1-v2)^2) where
L1=91.37 (@90%) 19.64(@20%)
L2=91.37 (@90%) 19.64(@20%)
u1= 0
u2= -8.392 (@90%), -1.803 (@20%)
v1= 0
v2= 13.64 (@90%) 2.93 (@20%)

This is a literal application of the CIE standard, and yet it gives an incorrect result: @ 20% x0.314, y0.351 is a good grayscale result. It isn't. A word to anyone else following this: If a calibrator tells you that this is an acceptable result, you should ask for your money back.

The correct implementation is the same formula, but with these values:
L1=100
L2=100
u1= 0
u2= -9.184
v1=0
v2= 14.92

for a dE of 17.5 regardless of the level of stimulus.

Bill, I understand what you are doing. That's the problem.

The rest of this is simply demonstrating that despite having published a dE calculator of your own, you do not really understand what goes on in the numbers. Your complaint, in this case, is with the CIE itself, not us..

Yep. The numbers are what they are, it's not our method. In fact, it is a lot closer to yours. The only thing I did was let L* vary, rather than hold it constant to show how the CIE intended the equation to work. Obviously, you did not recognize that.

If you think it is as easy to see a given color difference when there is approx. 1/7th - 1/10th as much light around as what you are adapted to for white then that sounds like a testable hypothesis. One might even develop a model, of people's color perception. Here's a hint: if you want to be taken seriously, don't use your "sunlight/inside light" example from your WSR article. That one was simply bad.This is just another repetition of what you have already said many times now. I invite anyone who cares to try to adjust their RGB biases to achieve x0.314, y0.351 @ 20% and see if you are able to clearly see the difference between that and x0.3127, y0.329 in test patterns and real program material.

What consequences are there for practical application? That you set a high hurdle for imperceptibility? That you have to rely on software that does all of the computation for you in real time? That our users have to toggle a control if they want to use it? Do tell. I'm curious here, too.

As for common sense: I'm all ears. I'd really be entertained to learn why you prefer tool vendors who were fast-and-loose with published standards.Again, I am not personally doing anything. Again, can you tell me of any other reputable source that uses the method you advocate for calculating dE for grayscale? Just one? Oh, yes, I already asked that question. I'm still waiting for an answer.

Careful, you are perilously close to saying that SMPTE uses L* when computing dE (you would be right there, though individuals vary in interpretation as this thread demonstrates). However, your math is off. Let's leave aside Lab76 alone until we master Luv. In CIE76 or CIE94 terms, it is over 9.That's odd, because you posted the number yourself as 18.4. But I guess that IS over 9. For the record, this is a REALLY minor point, but one you must emphasize as part of your "Tom doesn't understand" approach. The SMPTE standard is 4.0 Lab. That is roughly equivalent to 5.0 Luv. The dE of x0.314, y0.351 relative to Rec. 709 is 17.5 in CIELUV [dE(uv)] and 13.2 CIELAB in [dE(ab)].

Agreed. However, your "simple experiments", judging by the WSR article, often involve such "minor nuisances" as significant changes in ambient light spectrum and luminous flux, so one does have to double-check these things.You "agree" that the difference between 0.314, 0.351 and 0.3127, 0.329 at a typical 20% stim would be clearly visible and yet you continue to insist on a dE metric that reports the difference as perceptually negligible? OK, well that was easy. You just conceded the entire debate. BTW, is "ambient light spectrum and luminous flux" just a incredibly pretentious way of saying viewing conditions and perceived brightness? If you want to make a point here, you might want to fill that out a little.

Huh? It does, and so does yours, unless you use dE(uv) for gamut work. We recommend people turn off gamma correction when doing primary/secondary work. This was discussed previously.

A given absolute difference in chromaticity would be less visible at lower light levels. Whether it drops below the threshold of visibility is another matter. This is curious from someone who just published something on the underappreciated nature of brightness in color measurement. Ironic, even.You have basically 2 affirmative arguments for this methodology buried in the endless sniping at me: 1) it's the CIE standard; 2) people are less sensitive to color error at lower light levels so the dE number should reflect this. I was responding to 2), but on reflection I am not sure that this argument was sound. I withdraw it. BTW, this is the 3rd time you've referred to my WSR article. I'd be happy to comment on your published work in this area if you have any.

Huh? Are you confusing white, which is a point, with gray, which is a line connecting white and black? If white changes, then the entire adaptation model changes (must be re-normalized). Luminous flux is captured in the "Lightness" component of color error. You really don't understand this, do you?Whoa, are you actually insinuating that I don't understand something? Astonishing. You really need to try out some new material. No, Bill, I don't lack an understanding of white vs. gray.

Gauntlet thrown; gauntlet accepted. I am now free to post my critique of your CMS document. Hilarity will ensue.If this is typical of the level of discourse we can expect, I doubt that. You are, of course, free to post anything you like. It's a free country.

I'm not going to respond to your comments about Grassman's Law as I withdrew those comments in the next post, so it is a moot point.

Resurrecting an old thread.

I am using Delta E 1976 for greyscale. And for colour, Delta E 1994.

In Calman the default is 1976. That is what I used when calibrating the grayscale.

However, where it gets confusing is the colour tab. In Calman there is a box in the middle labelled Delta E 94 (with the readouts for luminance, saturation, hue) already regardless of what is set in the global options. My formula preference in the Calman options are still set to 1976. Do I need to change this option, is Calman being clever and over-riding it for this screen tab anyway? I realised after calibrating and getting everything to less than a Delta E of 1, that I had not changed the global setting because I was reading the Delta E 94 box on the tab and assumed I didn't need to do anything!

Resurrecting an old thread.

I am using Delta E 1976 for greyscale. And for colour, Delta E 1994.

In Calman the default is 1976. That is what I used when calibrating the grayscale.

However, where it gets confusing is the colour tab. In Calman there is a box in the middle labelled Delta E 94 (with the readouts for luminance, saturation, hue) already regardless of what is set in the global options. My formula preference in the Calman options are still set to 1976. Do I need to change this option, is Calman being clever and over-riding it for this screen tab anyway? I realised after calibrating and getting everything to less than a Delta E of 1, that I had not changed the global setting because I was reading the Delta E 94 box on the tab and assumed I didn't need to do anything!

I thought this would have a quick answer. But as I didn't, I did the experiment myself. I can confirm that no matter what you set as the global option for Delta E measurement, for the Colour tab, it forces the display/use of 1994 anyway. The global setting does not affect the resultant reading.

I thought this would have a quick answer. But as I didn't, I did the experiment myself. I can confirm that no matter what you set as the global option for Delta E measurement, for the Colour tab, it forces the display/use of 1994 anyway. The global setting does not affect the resultant reading.
This actually depends upon the layout. You can set up the layout to use the global option or a specific set. The default-delivered layout uses some elements of dE94. Probably a better topic for the SpectraCal support forum, though.

This actually depends upon the layout. You can set up the layout to use the global option or a specific set. The default-delivered layout uses some elements of dE94. Probably a better topic for the SpectraCal support forum, though.

That makes absolute sense. Thank you. Understood about the Spectracal forum :) If I have further questions I will ask them there.