Screen Printing HSB and Color Separations

[ScreenFoo]

Interesting stuff.  Thanks so much for chiming in, Mr. Coudray.  I didn't expect to see such an interesting connection drawn on this topic.

The impressionists had some rather interesting techniques with complements--as I understand it, the pure (unmixed) colors are what enabled their realistic interplay of light and shadow, as well as avoiding the pitfalls of the subtractive color model he was (and we are) stuck with.

If you look closely at something like A Sunday on Le Grande Jatte, you don't see a whole lot of black anywhere besides black clothes and the black dog, but the shadows are mostly complementary colors, as opposed to black.

I think It would be extremely interesting to see what an HSB style of sep does to a piece like that.

[AdvancedArtist]

Interesting stuff.  Thanks so much for chiming in, Mr. Coudray.  I didn't expect to see such an interesting connection drawn on this topic.

The impressionists had some rather interesting techniques with complements--as I understand it, the pure (unmixed) colors are what enabled their realistic interplay of light and shadow, as well as avoiding the pitfalls of the subtractive color model he was (and we are) stuck with.

If you look closely at something like A Sunday on Le Grande Jatte, you don't see a whole lot of black anywhere besides black clothes and the black dog, but the shadows are mostly complementary colors, as opposed to black.

I think It would be extremely interesting to see what an HSB style of sep does to a piece like that.

Interesting which one of these digital representations is the closest?

http://www.google.com/search?hl=en&site=imghp&tbm=isch&source=hp&biw=1120&bih=530&q=A+Sunday+on+Le+Grande+Jatte&oq=A+Sunday+on+Le+Grande+Jatte&gs_l=img.12..0i10i24.1906.1906.0.3509.1.1.0.0.0.0.210.210.2-1.1.0...0.0...1ac.1.9.img.b16Lh2OiHt8

I am stumped....

Edit.. its like skittles that dress in every color in the rainbow. Purple, blue, brown, green etc.

[blue moon]

Pierre asked me to take a look at this discussion and add my two cents worth.  For me, it's a trip back in time. We wrestled with these issues extensively in '94-'95 and the realization that Photoshop would never be able to address the core underlying separation issues dealing with chromatic and achromatic gray balance. This is why ICISS was developed.  Photoshop does not possess the core engine to address nChannel separations where n is the number of Channels and each channel represents THE SPECIFIC COLOR COORDINANTS OF THE PRINTING INK.

Where I see separation models failing, and it is happens all the time, is where the nChannel color position is established based on a numerical value that IS NOT representative of the ink being printed. This is one of the primary causes for color shifting on press compared to the sep. The more saturated and pure the color used for the channel separation, the greater the mismatch.  So, with that being said, I would like to spend a couple of minutes on the points Pierre was making about using 187 as a primary or adding black to 185 to make the 187.

First a bit of color theory, all based on HSB because it is definitely the most understandable and intuituve of the color spaces.  Every color has the theretical possibility of having 100% Brightness and 100% Saturation. That is theory.  In reality, we cannot achieve this for two reasons. The first is that our visual sensory perception does not allow us to render full satuation at very high or very low Brightness levels.  As the Brightness drops, the saturations moves closer and closer to black until we are no longer able to perceive the color and we only see black.  The same is true for very high Brightness values as we move toward white. 

It is important to note that this is NONLINEAR meaning that every color has a different perception profile (how we perceive the color in our brains.)  This is what makes the math so incredibly difficult to calculate. It is essentiall 3D Matrix Calculus where we have a different function for every hue angle. In ICISS we set it up to measure and calculate to the first decimal place or 3600 different color transforms that interact in 3D. It was so involved 2 PhD mathematicians were brought in as consultants to interperet the actual printing data into an algorithm that would accurately render the complexity.  I only bring this up to illustrate where the discussion eventually led to the realization this wasn't going to be solved with the tools inside PS.

Now back to the discussion using Pierre's example.  If you cannot achieve a color with a natural pigment, you have to find a way to darken the color and still preserve as much of the color as possible.  It will be easier to understand if you define "muddiness" as the addition of gray or desaturation of color to gray.  Since we cannot preserve the saturation, any addition of gray will desaturate the color, adding grayness or muddiness to the color.

The two ways to desaturate are chromatically and achromatically. Chromatic desaturation is the addition of the complemenatry color or the mixing of a Hue 180° opposed to the color you are mixing into it.  So when Pierre is Using Blue Shade Red and Yellow Shade Red to get his 187, what he is really doing is combining the Blue Shade component and the Yellow Shade component (complements eg 60° and 240° ) to make a chromatic gray that darkens the Red.   This is the simple explanation since the amount of shift off the pure Hue is different for the BS Red than it is for the YS Red. A colorimeter reading of each pigment is necessary to determine exactly how much of a shift has occurred for each of the pigments based on the color strength or density of each pigment. The darker the pigment, the more blue emphasis due to the physics of light scattter.

The other way to do this is Achromatically (a = Without, chromatic = color). This means replacing the chromatic complement with black.  Essentially simplifying the equation by replacing two components with one.  This has the same darkening and desaturating effect but with one big difference.  As any artist will tell you, when you add black to any color you KILL the chromatic aspect of that color.  So it is ALWAYS preferable to desaturate with the chromatic complement than to add black.

From a printing aspect the replacement of the complementatry color with black is called GCR. Gray Component Replacement where the Gray Component Replacement is Black. It is a fantastic tool for stabilizing tone shifting on press, but at the expense of color saturation, especially in the darker colors. Photoshop uses GCR on a wholesale basis, applying it to all colors universally.  When you understand how all this works it's possible to design selective GCR for every Hue angle.  Since it is just math and it's an algorithm with variables, we just define a variable for the coefficient of chromatic > achromatic replacement.  This means we can have complementary replacement  for a color like Yellow where the addition of black is destructive to the color ( Yellow + Black = perceived muddy green), or we can have 100% black replacement in dark blues and reds where the addition of black is less destructive to the color. We can also decide if we want to use achromatic black in neutral colors and chromatic black for darker colors, as well was where the transition will occur between light and dark.  All in all, a lot to think about and even more to model and calculate. It's like a rabbit hole, once you go down it, you keep on going.

Where all this becomes super important are in the pastel colors.  We know this most commonly as the complex natural fleshtones.  The addition of black in pastel colors is EXTREMELY destructive and desaturating to the color.  This is were muddiness is exaggerated ten fold (literally because our eye is at least 10x more sensitive to tone and color changes in pastels.) This is also the reason almost all atttempts at rendering natural complex fleshtone fail.  Anyone who has tried to deal with this or florals, animal fur (chromatic grays,) early sunrises, or pastel neutrals like beach sand, etc. knows this well.

Color is so much like learning to appreciate fine wine. In the beginning you can tell the difference between the broad categories white and red.  You will typically prefer sweeter wines (saturation differentation).  As you mature and develop your sensory skills you will be able to differentiate the varieties in each area (hue differentation.) Ultimately you will be able to detect the subtile differences between lightness  (whites and rosé) and heavy (reds) [Brightness.]  The very, very fine diffenences we often joke about (hints of tobacco, vanilla, plum, and razzberry etc) are really the super refined percepts of combining all elements together.

In color, it is not just the numerics.  It is perceptual. Things like simultaneous contrast and nonlinear tone sensitivity are extremely difficult to model and add yet another dimension to the separation element.  And then their is the psychological influence of color on mood that shifts the perception independently of the math (I was so mad I saw Red, or I'm feeling a little Blue today.)  There is a reason color is used in advertising and architecture to effect mood. It all comes into play.

So where does all this leave us?  Essentially the more tools we can build in as addressible variables, the better we can interpret color based on intended audience and the intended viewing environment.  There is no universal solution, only the tools to interpret.  High end color will always be edited. You cannot standardize it. Standardization = commoditization. So Dan, you are safe and will always be safe in this arena.

When you look at a true master like Andy Anderson, you realize this is a neverending quest for the ultimate reproduction. When you seek to render the transluscent luminence of a youthful skintone or the subtle reflection off a metallic or transparent color surface, you realize just how amazing our brain is at interpreting color and why any of us who have spent our lives in search of the answers will never be satisfied in our quest.

few thoughts and questions. . .

As I have not done any scientific research and poses no relevant data (short of very subjective observations), anything contributed will have to be an opinion or a question. Mark is the one with hard facts here and I am willing to take his word as gospel until there is an opportunity to do the research and get data that contradicts what he is saying (which is not very likely to happen. The contradicting data part that is).
Reading this post, the first question I have is about the blue emphasis with the darker pigments. Mark, is there a name for this effect and where do we get more info?

When you say, there is an achromatic and chromatic variable (specifying the point of shifting from a to chromatic and vice versa) are you talking about your software or Photoshop? I know you can specify the black generation in the CMYK settings, but can this be done for other colorspace models? And while at it, what is a dark yellow anyways? It seems, no matter what is done to it, it always shifts to one of its neighbors, green or red.

what about using out of gamut primaries for separations? 240/100/100 is showing as out of range in Photoshop, but is that the limitation of the colorspace, physics or the software? If 240/100/100 is not achievable, then the only solution to correct separations is the nChannel model!

In color, it is not just the numerics.  It is perceptual

.
At what point could it be said that it is not perceptual any more? It seems to me that ultimately there are 1's and 0's and that is the art we are dealing with today. It is specified very precisely, and just because somebody perceives it differently it does not change the underlying data. It seems to me that as deep as this rabbit hole is, eventually there is a hard bottom (linear system assumption). Once down at the 0's and 1's level, there is no more interpreting. The data is not open to individual perceptions if it is manipulated by math rather than subjective opinions (since math has objective rules only). What is it going to take to get to that point? Are we all gonna have to work in the same colorspace that is independent of the media? Is peeling back all these layers of interference the answer to getting the true result? Is that even a possibility or is the model non linear and all the results are just approximations? If the model is linear, then there is an answer and any deviation is an interpretation, either personal or by request, but an interpretation none the less.

time to go though, I'll have to tackle the second part of this tomorrow. . .

pierre

[AdvancedArtist]

The water is deep and warm. DIVE IN!


Wow. Great stuff. Good thread. For now, I don't have much to contribute.

You have allot to contribute Dan decades of experience and well you know. You are DTD I may be strong about trying to get this out but I have always respected your work.  Sometimes it may not come across that way.. but getting these things out is very important.

[AdvancedArtist]

Pierre asked me to take a look at this discussion and add my two cents worth.  For me, it's a trip back in time. We wrestled with these issues extensively in '94-'95 and the realization that Photoshop would never be able to address the core underlying separation issues dealing with chromatic and achromatic gray balance. This is why ICISS was developed.  Photoshop does not possess the core engine to address nChannel separations where n is the number of Channels and each channel represents THE SPECIFIC COLOR COORDINANTS OF THE PRINTING INK.

Where I see separation models failing, and it is happens all the time, is where the nChannel color position is established based on a numerical value that IS NOT representative of the ink being printed. This is one of the primary causes for color shifting on press compared to the sep. The more saturated and pure the color used for the channel separation, the greater the mismatch.  So, with that being said, I would like to spend a couple of minutes on the points Pierre was making about using 187 as a primary or adding black to 185 to make the 187.

First a bit of color theory, all based on HSB because it is definitely the most understandable and intuituve of the color spaces.  Every color has the theretical possibility of having 100% Brightness and 100% Saturation. That is theory.  In reality, we cannot achieve this for two reasons. The first is that our visual sensory perception does not allow us to render full satuation at very high or very low Brightness levels.  As the Brightness drops, the saturations moves closer and closer to black until we are no longer able to perceive the color and we only see black.  The same is true for very high Brightness values as we move toward white. 

It is important to note that this is NONLINEAR meaning that every color has a different perception profile (how we perceive the color in our brains.)  This is what makes the math so incredibly difficult to calculate. It is essentiall 3D Matrix Calculus where we have a different function for every hue angle. In ICISS we set it up to measure and calculate to the first decimal place or 3600 different color transforms that interact in 3D. It was so involved 2 PhD mathematicians were brought in as consultants to interperet the actual printing data into an algorithm that would accurately render the complexity.  I only bring this up to illustrate where the discussion eventually led to the realization this wasn't going to be solved with the tools inside PS.

Now back to the discussion using Pierre's example.  If you cannot achieve a color with a natural pigment, you have to find a way to darken the color and still preserve as much of the color as possible.  It will be easier to understand if you define "muddiness" as the addition of gray or desaturation of color to gray.  Since we cannot preserve the saturation, any addition of gray will desaturate the color, adding grayness or muddiness to the color.

The two ways to desaturate are chromatically and achromatically. Chromatic desaturation is the addition of the complemenatry color or the mixing of a Hue 180° opposed to the color you are mixing into it.  So when Pierre is Using Blue Shade Red and Yellow Shade Red to get his 187, what he is really doing is combining the Blue Shade component and the Yellow Shade component (complements eg 60° and 240° ) to make a chromatic gray that darkens the Red.   This is the simple explanation since the amount of shift off the pure Hue is different for the BS Red than it is for the YS Red. A colorimeter reading of each pigment is necessary to determine exactly how much of a shift has occurred for each of the pigments based on the color strength or density of each pigment. The darker the pigment, the more blue emphasis due to the physics of light scattter.

The other way to do this is Achromatically (a = Without, chromatic = color). This means replacing the chromatic complement with black.  Essentially simplifying the equation by replacing two components with one.  This has the same darkening and desaturating effect but with one big difference.  As any artist will tell you, when you add black to any color you KILL the chromatic aspect of that color.  So it is ALWAYS preferable to desaturate with the chromatic complement than to add black.

From a printing aspect the replacement of the complementatry color with black is called GCR. Gray Component Replacement where the Gray Component Replacement is Black. It is a fantastic tool for stabilizing tone shifting on press, but at the expense of color saturation, especially in the darker colors. Photoshop uses GCR on a wholesale basis, applying it to all colors universally.  When you understand how all this works it's possible to design selective GCR for every Hue angle.  Since it is just math and it's an algorithm with variables, we just define a variable for the coefficient of chromatic > achromatic replacement.  This means we can have complementary replacement  for a color like Yellow where the addition of black is destructive to the color ( Yellow + Black = perceived muddy green), or we can have 100% black replacement in dark blues and reds where the addition of black is less destructive to the color. We can also decide if we want to use achromatic black in neutral colors and chromatic black for darker colors, as well was where the transition will occur between light and dark.  All in all, a lot to think about and even more to model and calculate. It's like a rabbit hole, once you go down it, you keep on going.

Where all this becomes super important are in the pastel colors.  We know this most commonly as the complex natural fleshtones.  The addition of black in pastel colors is EXTREMELY destructive and desaturating to the color.  This is were muddiness is exaggerated ten fold (literally because our eye is at least 10x more sensitive to tone and color changes in pastels.) This is also the reason almost all atttempts at rendering natural complex fleshtone fail.  Anyone who has tried to deal with this or florals, animal fur (chromatic grays,) early sunrises, or pastel neutrals like beach sand, etc. knows this well.

Color is so much like learning to appreciate fine wine. In the beginning you can tell the difference between the broad categories white and red.  You will typically prefer sweeter wines (saturation differentation).  As you mature and develop your sensory skills you will be able to differentiate the varieties in each area (hue differentation.) Ultimately you will be able to detect the subtile differences between lightness  (whites and rosé) and heavy (reds) [Brightness.]  The very, very fine diffenences we often joke about (hints of tobacco, vanilla, plum, and razzberry etc) are really the super refined percepts of combining all elements together.

In color, it is not just the numerics.  It is perceptual. Things like simultaneous contrast and nonlinear tone sensitivity are extremely difficult to model and add yet another dimension to the separation element.  And then their is the psychological influence of color on mood that shifts the perception independently of the math (I was so mad I saw Red, or I'm feeling a little Blue today.)  There is a reason color is used in advertising and architecture to effect mood. It all comes into play.

So where does all this leave us?  Essentially the more tools we can build in as addressible variables, the better we can interpret color based on intended audience and the intended viewing environment.  There is no universal solution, only the tools to interpret.  High end color will always be edited. You cannot standardize it. Standardization = commoditization. So Dan, you are safe and will always be safe in this arena.

When you look at a true master like Andy Anderson, you realize this is a neverending quest for the ultimate reproduction. When you seek to render the transluscent luminence of a youthful skintone or the subtle reflection off a metallic or transparent color surface, you realize just how amazing our brain is at interpreting color and why any of us who have spent our lives in search of the answers will never be satisfied in our quest.

few thoughts and questions. . .

As I have not done any scientific research and poses no relevant data (short of very subjective observations), anything contributed will have to be an opinion or a question. Mark is the one with hard facts here and I am willing to take his word as gospel until there is an opportunity to do the research and get data that contradicts what he is saying (which is not very likely to happen. The contradicting data part that is).
Reading this post, the first question I have is about the blue emphasis with the darker pigments. Mark, is there a name for this effect and where do we get more info?

When you say, there is an achromatic and chromatic variable (specifying the point of shifting from a to chromatic and vice versa) are you talking about your software or Photoshop? I know you can specify the black generation in the CMYK settings, but can this be done for other colorspace models? And while at it, what is a dark yellow anyways? It seems, no matter what is done to it, it always shifts to one of its neighbors, green or red.

what about using out of gamut primaries for separations? 240/100/100 is showing as out of range in Photoshop, but is that the limitation of the colorspace, physics or the software? If 240/100/100 is not achievable, then the only solution to correct separations is the nChannel model!

In color, it is not just the numerics.  It is perceptual

.
At what point could it be said that it is not perceptual any more? It seems to me that ultimately there are 1's and 0's and that is the art we are dealing with today. It is specified very precisely, and just because somebody perceives it differently it does not change the underlying data. It seems to me that as deep as this rabbit hole is, eventually there is a hard bottom (linear system assumption). Once down at the 0's and 1's level, there is no more interpreting. The data is not open to individual perceptions if it is manipulated by math rather than subjective opinions (since math has objective rules only). What is it going to take to get to that point? Are we all gonna have to work in the same colorspace that is independent of the media? Is peeling back all these layers of interference the answer to getting the true result? Is that even a possibility or is the model non linear and all the results are just approximations? If the model is linear, then there is an answer and any deviation is an interpretation, either personal or by request, but an interpretation none the less.

time to go though, I'll have to tackle the second part of this tomorrow. . .

pierre

I think Jeffs test pattern/color wheel would be a great way to start it has a good representation of many hues/colors all kinds with saturation and brightness. We have not printed it yet but I think that or something close to it would be the best graphic to make indepth analysis with. Maybe in a 6 hues and a 12 tints pull which would give us some data relating to color quality and vibrancy based on number colors and saturation/brightness across a number of blends and colors.

[blue moon]

Pierre asked me to take a look at this discussion and add my two cents worth.  For me, it's a trip back in time. We wrestled with these issues extensively in '94-'95 and the realization that Photoshop would never be able to address the core underlying separation issues dealing with chromatic and achromatic gray balance. This is why ICISS was developed.  Photoshop does not possess the core engine to address nChannel separations where n is the number of Channels and each channel represents THE SPECIFIC COLOR COORDINANTS OF THE PRINTING INK.

Where I see separation models failing, and it is happens all the time, is where the nChannel color position is established based on a numerical value that IS NOT representative of the ink being printed. This is one of the primary causes for color shifting on press compared to the sep. The more saturated and pure the color used for the channel separation, the greater the mismatch.  So, with that being said, I would like to spend a couple of minutes on the points Pierre was making about using 187 as a primary or adding black to 185 to make the 187.

First a bit of color theory, all based on HSB because it is definitely the most understandable and intuituve of the color spaces.  Every color has the theretical possibility of having 100% Brightness and 100% Saturation. That is theory.  In reality, we cannot achieve this for two reasons. The first is that our visual sensory perception does not allow us to render full satuation at very high or very low Brightness levels.  As the Brightness drops, the saturations moves closer and closer to black until we are no longer able to perceive the color and we only see black.  The same is true for very high Brightness values as we move toward white. 

It is important to note that this is NONLINEAR meaning that every color has a different perception profile (how we perceive the color in our brains.)  This is what makes the math so incredibly difficult to calculate. It is essentiall 3D Matrix Calculus where we have a different function for every hue angle. In ICISS we set it up to measure and calculate to the first decimal place or 3600 different color transforms that interact in 3D. It was so involved 2 PhD mathematicians were brought in as consultants to interperet the actual printing data into an algorithm that would accurately render the complexity.  I only bring this up to illustrate where the discussion eventually led to the realization this wasn't going to be solved with the tools inside PS.

Now back to the discussion using Pierre's example.  If you cannot achieve a color with a natural pigment, you have to find a way to darken the color and still preserve as much of the color as possible.  It will be easier to understand if you define "muddiness" as the addition of gray or desaturation of color to gray.  Since we cannot preserve the saturation, any addition of gray will desaturate the color, adding grayness or muddiness to the color.

The two ways to desaturate are chromatically and achromatically. Chromatic desaturation is the addition of the complemenatry color or the mixing of a Hue 180° opposed to the color you are mixing into it.  So when Pierre is Using Blue Shade Red and Yellow Shade Red to get his 187, what he is really doing is combining the Blue Shade component and the Yellow Shade component (complements eg 60° and 240° ) to make a chromatic gray that darkens the Red.   This is the simple explanation since the amount of shift off the pure Hue is different for the BS Red than it is for the YS Red. A colorimeter reading of each pigment is necessary to determine exactly how much of a shift has occurred for each of the pigments based on the color strength or density of each pigment. The darker the pigment, the more blue emphasis due to the physics of light scattter.

The other way to do this is Achromatically (a = Without, chromatic = color). This means replacing the chromatic complement with black.  Essentially simplifying the equation by replacing two components with one.  This has the same darkening and desaturating effect but with one big difference.  As any artist will tell you, when you add black to any color you KILL the chromatic aspect of that color.  So it is ALWAYS preferable to desaturate with the chromatic complement than to add black.

From a printing aspect the replacement of the complementatry color with black is called GCR. Gray Component Replacement where the Gray Component Replacement is Black. It is a fantastic tool for stabilizing tone shifting on press, but at the expense of color saturation, especially in the darker colors. Photoshop uses GCR on a wholesale basis, applying it to all colors universally.  When you understand how all this works it's possible to design selective GCR for every Hue angle.  Since it is just math and it's an algorithm with variables, we just define a variable for the coefficient of chromatic > achromatic replacement.  This means we can have complementary replacement  for a color like Yellow where the addition of black is destructive to the color ( Yellow + Black = perceived muddy green), or we can have 100% black replacement in dark blues and reds where the addition of black is less destructive to the color. We can also decide if we want to use achromatic black in neutral colors and chromatic black for darker colors, as well was where the transition will occur between light and dark.  All in all, a lot to think about and even more to model and calculate. It's like a rabbit hole, once you go down it, you keep on going.

Where all this becomes super important are in the pastel colors.  We know this most commonly as the complex natural fleshtones.  The addition of black in pastel colors is EXTREMELY destructive and desaturating to the color.  This is were muddiness is exaggerated ten fold (literally because our eye is at least 10x more sensitive to tone and color changes in pastels.) This is also the reason almost all atttempts at rendering natural complex fleshtone fail.  Anyone who has tried to deal with this or florals, animal fur (chromatic grays,) early sunrises, or pastel neutrals like beach sand, etc. knows this well.

Color is so much like learning to appreciate fine wine. In the beginning you can tell the difference between the broad categories white and red.  You will typically prefer sweeter wines (saturation differentation).  As you mature and develop your sensory skills you will be able to differentiate the varieties in each area (hue differentation.) Ultimately you will be able to detect the subtile differences between lightness  (whites and rosé) and heavy (reds) [Brightness.]  The very, very fine diffenences we often joke about (hints of tobacco, vanilla, plum, and razzberry etc) are really the super refined percepts of combining all elements together.

In color, it is not just the numerics.  It is perceptual. Things like simultaneous contrast and nonlinear tone sensitivity are extremely difficult to model and add yet another dimension to the separation element.  And then their is the psychological influence of color on mood that shifts the perception independently of the math (I was so mad I saw Red, or I'm feeling a little Blue today.)  There is a reason color is used in advertising and architecture to effect mood. It all comes into play.

So where does all this leave us?  Essentially the more tools we can build in as addressible variables, the better we can interpret color based on intended audience and the intended viewing environment.  There is no universal solution, only the tools to interpret.  High end color will always be edited. You cannot standardize it. Standardization = commoditization. So Dan, you are safe and will always be safe in this arena.

When you look at a true master like Andy Anderson, you realize this is a neverending quest for the ultimate reproduction. When you seek to render the transluscent luminence of a youthful skintone or the subtle reflection off a metallic or transparent color surface, you realize just how amazing our brain is at interpreting color and why any of us who have spent our lives in search of the answers will never be satisfied in our quest.

few thoughts and questions. . .

As I have not done any scientific research and poses no relevant data (short of very subjective observations), anything contributed will have to be an opinion or a question. Mark is the one with hard facts here and I am willing to take his word as gospel until there is an opportunity to do the research and get data that contradicts what he is saying (which is not very likely to happen. The contradicting data part that is).
Reading this post, the first question I have is about the blue emphasis with the darker pigments. Mark, is there a name for this effect and where do we get more info?

When you say, there is an achromatic and chromatic variable (specifying the point of shifting from a to chromatic and vice versa) are you talking about your software or Photoshop? I know you can specify the black generation in the CMYK settings, but can this be done for other colorspace models? And while at it, what is a dark yellow anyways? It seems, no matter what is done to it, it always shifts to one of its neighbors, green or red.

what about using out of gamut primaries for separations? 240/100/100 is showing as out of range in Photoshop, but is that the limitation of the colorspace, physics or the software? If 240/100/100 is not achievable, then the only solution to correct separations is the nChannel model!

In color, it is not just the numerics.  It is perceptual

.
At what point could it be said that it is not perceptual any more? It seems to me that ultimately there are 1's and 0's and that is the art we are dealing with today. It is specified very precisely, and just because somebody perceives it differently it does not change the underlying data. It seems to me that as deep as this rabbit hole is, eventually there is a hard bottom (linear system assumption). Once down at the 0's and 1's level, there is no more interpreting. The data is not open to individual perceptions if it is manipulated by math rather than subjective opinions (since math has objective rules only). What is it going to take to get to that point? Are we all gonna have to work in the same colorspace that is independent of the media? Is peeling back all these layers of interference the answer to getting the true result? Is that even a possibility or is the model non linear and all the results are just approximations? If the model is linear, then there is an answer and any deviation is an interpretation, either personal or by request, but an interpretation none the less.

time to go though, I'll have to tackle the second part of this tomorrow. . .

pierre

I think Jeffs test pattern/color wheel would be a great way to start it has a good representation of many hues/colors all kinds with saturation and brightness. We have not printed it yet but I think that or something close to it would be the best graphic to make indepth analysis with. Maybe in a 6 hues and a 12 tints pull which would give us some data relating to color quality and vibrancy based on number colors and saturation/brightness across a number of blends and colors.

email me the seps and I'll print it and take the spectrometer readings. While the color wheel would be nice to have, it would make it hard to take the readings. Make it blocks of at least 1x1" (and 2x2 would be better). let's say 12 wide and 14 tall.

pierre

p.s. I have no idea what kind of readings we'll get from the garment, but it should be worth a try.

[AdvancedArtist]

Let me get the file from Jeff I think some of detail may be to small in his current version but we can make a file that will basically give us the same info with bigger elements making it easier to examine the results. I will go over this with him and then get with you.

[drdot]

few thoughts and questions. . .

As I have not done any scientific research and poses no relevant data (short of very subjective observations), anything contributed will have to be an opinion or a question. Mark is the one with hard facts here and I am willing to take his word as gospel until there is an opportunity to do the research and get data that contradicts what he is saying (which is not very likely to happen. The contradicting data part that is).
Reading this post, the first question I have is about the blue emphasis with the darker pigments. Mark, is there a name for this effect and where do we get more info?

When you say, there is an achromatic and chromatic variable (specifying the point of shifting from a to chromatic and vice versa) are you talking about your software or Photoshop? I know you can specify the black generation in the CMYK settings, but can this be done for other colorspace models? And while at it, what is a dark yellow anyways? It seems, no matter what is done to it, it always shifts to one of its neighbors, green or red.

what about using out of gamut primaries for separations? 240/100/100 is showing as out of range in Photoshop, but is that the limitation of the colorspace, physics or the software? If 240/100/100 is not achievable, then the only solution to correct separations is the nChannel model!

Chromatic gray can be made from any combination of colors. The definition of gray is R=G=B value, so a 50% neutral gray is 128, 128, 128.  This is perfect neutral gray.  If you have two colors that are numerically (H) separated by 180° you will get a perfect neutral gray (NG.)  With the right mixing algorithm, you can make neutral gray from any 3 colors, or as many as you like for that matter.  ICISS has the ability to make from any combination of colors and to transition to Black (Achromatic Gray) at any point.  Most of you will be familiar with this concept in Photoshop>Color Settings>Custom CMYK>Black Generation.  When set to Medium, you will note that the Black (here chromatic) is being made from pure CMYK until about the 25% tonal value when Black makes an appearance and increases progressively.  The color component maxes out at 50% tonal gray and the Black continues to add in order to get the maximum full tonal gray scale.

Pay particular attention to the differences between the Cyan, Magenta, and Yellow values to make neutral gray.  This demonstrates that we need less Yellow because each of the other colors has Yellow containation. The Magenta has up to 50% yellow shift from the ideal Magenta math point (300°)  It also has some Cyan contaimination as well (9%- 11% typically.)  The contamination of color in a primay is referred to as the "grayness" of the color and represents the natural complementary protion in the ink.  For a  typical Magenta the Grayness  is  9%- 11% and the Hue Shift is 30%. 

This is how a set of process colors  positions is analyzed and it is what I used when I worked with Terry Kaiserman in 1991 to develop the Union TruTone Process ink set (still the best process ink set by far).

In color, it is not just the numerics.  It is perceptual.

At what point could it be said that it is not perceptual any more? It seems to me that ultimately there are 1's and 0's and that is the art we are dealing with today. It is specified very precisely, and just because somebody perceives it differently it does not change the underlying data. It seems to me that as deep as this rabbit hole is, eventually there is a hard bottom (linear system assumption). Once down at the 0's and 1's level, there is no more interpreting. The data is not open to individual perceptions if it is manipulated by math rather than subjective opinions (since math has objective rules only). What is it going to take to get to that point? Are we all gonna have to work in the same colorspace that is independent of the media? Is peeling back all these layers of interference the answer to getting the true result? Is that even a possibility or is the model non linear and all the results are just approximations? If the model is linear, then there is an answer and any deviation is an interpretation, either personal or by request, but an interpretation none the less.

pierre

This is like saying, "If a tree falls in the forest, and there is no one there, is there still sound?"  The answer is, of course there is sound.  Perception is our ability to detect the presence of color. Perception changes under the view conditions, age and gender of the viewer, genetics, and a host of other factors.  Look at the far right side of the color picker in PS. As any Hue of 100% S moves from from B = 100 to B=0%, at some point on YOUR monitor under YOUR viewing conditions YOU will no longer be able to perceive or detect color, even though the VERY low B% appears to be BLACK.  This is one of the reasons HSB pulls such accurate seps. It is going for the color values and is delivering color information into the appropriate ink channel.  Whether we can reproduce that color is dependent on the very best pigment we have to work with.  So if you want 240°, 100, 100  - the very best you're going to get from a Pantone Color is 2736  (234, 83, 66).  Any color with a value higher than the S or B value will be Out of Gamut. Interestingly, as you go up one step in the book to Pantone 2726, you will notice a HUE SHIFT to the Yellow Side of 4°, a DECREASE in Saturation and an INCREASE in Brightness.  This illustrates a the yellow shift I spoke of earlier because there  are less pigment molecules in a given space (lower density) and as a consequence, there is less light scatter and absorbsion resulting in more longer wavelength frequencies being reflected to our eye.  We see the same efffect when we look at water in swimming pool from above, or the shallow areas of a tropical beach. Shallow=yellow cast, Deep = Red cast.

[AdvancedArtist]

Tony,
Underbasing is another really complex idea. And once again, the solutions out there are all over the place.  Underbasing works best with the HSB model (surprise.) 

This is a profound statement concerning HSB because we screen printers print with white allot. Which is partially why HSB is the perfect fit for us. White and white bases are something we deal with all the time and HSB is perfectly suited for the ways in which we use white.

HSB easier and uniquely suited to us. Allot of technical posts here but do not let that confuse you before all this is done I will break it all down into much easier terms in video.

[artprofile]

If Mark can find some time, it would be great to have a post showing the capabilities and features of ICISS.

How and what exactly to do with the equipment we have is still a little unclear. This thread is definitely shaping a path that leads to understanding what is really going on. I could randomly start taking the readings of what colors are doing when printed, but without truly understanding what is going on, it would be like whacking around hoping to hit the piñata! Systematic and continuous advancement leads to progress, randomly flailing around just makes one tired!

pierre

pierre

... so very sorry that the topic is frozen ...I want to know the possibility of ISISS - from the Maestro!

[Frog]

If Mark can find some time, it would be great to have a post showing the capabilities and features of ICISS.

How and what exactly to do with the equipment we have is still a little unclear. This thread is definitely shaping a path that leads to understanding what is really going on. I could randomly start taking the readings of what colors are doing when printed, but without truly understanding what is going on, it would be like whacking around hoping to hit the piñata! Systematic and continuous advancement leads to progress, randomly flailing around just makes one tired!

pierre

pierre

... so very sorry that the topic is frozen ...I want to know the possibility of ISISS - from the Maestro!

Not frozen, just stalled. As was mentioned, the Doctor doesn't have as much spare time as many of us do.
Who knows? Perhaps a PM or email could get it going again and answer some specifics.

[artprofile]

Ok! I'll follow the theme. with hope!

[willy35]

Too bad this thread stops.

I have a question about iciss. 
Why is it so expensive,  why since 1994 no software can match its accuracy?
I had a demo,  very impressive,  you can extract colors so so easily,  and so so faaast... 

Is there lot of shop using it?  Is there any good software that can come close to it ? 

Thank you

[Dottonedan]

Mark Coudrey's separation program is and was THE separation program to use. No other sep program on the market today or previously has come close. Period.

Why the high price?  For those who can afford it, the price is of little concern for what it does.

Other sep programs are very useful for the average shop and you can tweak to perfection but Coudreys has always been the best.