Dan Margulis Applied Color Theory

Natural Greens

   Date: Wed, 1 Jun 2005 20:17:28 EDT
   From: Dan Margulis
Subject: Natural Greens

*********FORWARDED MESSAGE**********

Subj:   Natural greens
Date:   Wednesday, June 1, 2005 2:31:26 PM
From:   yelis@jet.msk.su
To:     Dan Margulis

Dan,

Sorry for sending message directly - I am not sure that my question is of
interest to the colortheory list.

In your book (4th edition) you say that greenish yellow is impossible for
natural greens (pg.27 and elsewhere). On the other hand on pg.216 (fig.
11.8) you discuss artichoke color - 74L (8)A 17B - and say that it is green
and could be right. Am I missing something - because I think that color is
greenish yellow and you say so in the table on pg.100? Does that mean that
greenish yellows are acceptable?

It is not a theoretical question for me - I was working on a file with lots
of greenish yellows (G slightly above R, B almost two times less) and good
neutrals (rocks, birch trunks, etc.) All my attempts to make greens green
rather than yellow made image significantly less believable.

Best regards,
Vladimir Yelisseyev
___________________________________________________________________________

   Date: Wed, 1 Jun 2005 22:12:47 -0400
   From: "Michael Demyan"
Subject: RE: Natural Greens

Hi Vlad:

Did you try using the Hue/Saturation tool? Select Greens from the drop down and then use the color picker+ to select a few of your "yellow-greens" and then adjust the hue slider to just affect the greens you selected. This approach has worked for me for other color adjustments.

Michael Demyan
Fine Photography & Digital Graphic Design
www.mikedemyan.com
___________________________________________________________________________

   Date: Thu, 02 Jun 2005 08:50:31 -0000
   From: "v_yelis"
Subject: Re: Natural Greens

First part of my question was to clarify - are such colors Ok? Because in Dan's book in some places (pg.27, etc.) they are stated as impossible, but then on pg.216 as Ok (and even after correction they remain yellow-greens).

Vladimir
___________________________________________________________________________

   Date: Thu, 2 Jun 2005 09:30:29 EDT
   From: Dan Margulis
Subject: Re: Natural Greens

Vladimir writes,

In your book (4th edition) you say that greenish yellow is impossible for
natural greens (pg.27 and elsewhere). On the other hand on pg.216 (fig.
11.8) you discuss artichoke color - 74L (8)A 17B - and say that it is green
and could be right. Am I missing something - because I think that color is
greenish yellow and you say so in the table on pg.100? Does that mean that
greenish yellows are acceptable?

In the LAB definition, yes, in the CMYK definition, no. When working with green colors in LAB, I recommend that you set the right half of the Info palette to read CMYK equivalents, because the rule is simpler to understand.

In CMYK, Photoshop defines "green" as cyan= yellow. This color is too blue for almost all plant life. Instead, the yellow must be higher, but not so much higher as to make the plant a greenish yellow rather than a yellowish green. The most extreme values normally found would be Y=C for a very blue green and Y=2C for a very yellow one.

In LAB, "green" (meaning A is negative, B is zero) is a much bluer color than CMYK "green". It is almost halfway to cyan. So the "green" in LAB is the color referred to in English as "teal."
No plant life can possibly have 0b, because that would convert to CMYK as Y<C. Even a very blue plant probably has a=2*(-b).

Because the range begins so far away from the "green" point, it goes well beyond the midpoint between "green" (A negative, B zero) and "yellow" (A zero, B positive). It is even possible to find a plant that is 3a=(-b).

The artichoke color, 74L(8)a17b, is therefore acceptable. It converts to CMYK as 33C18M46Y, a dull, yellowish green.

Dan Margulis
___________________________________________________________________________

   Date: Thu, 02 Jun 2005 19:36:48 -0000
   From: "v_yelis"
Subject: Re: Natural Greens

Thanks, Dan.
Vladimir
___________________________________________________________________________

   Date: Fri, 24 Jun 2005 14:43:02 -0500
   From: Jerry Sedgewick
Subject: Pure color green

Hello everyone,

I'm a newbie to this group and a scientist/imaging specialist who works almost exclusively with colorized grayscale images.  In this world I belong to, detectors (primarily photomultiplier tubes: PMTs) collect photons and then these are mapped to grayscale LUTs (Look Up Tables) or to colorized LUTs for display.  The user can acquire these images as TIFF files, with some manufacturers allowing the "save" to include the color LUT overlay (often a screen capture) as an indexed color image; or as an indexed color image with a grayscale LUT.

In all instances, the LUTs are set by default to colorize one channel only, either in red, green or blue.  When a mode change is done to RGB Color, the channel of interest is colorized and the other two channels are completely black.

You can imagine the nightmare, then, of printing these images while maintaining brightness in the green, red and (especially) blue channels because of gamut issues.

I have been able to change the RGB colors for the red and blue so that these are not pure and so that these colors reproduce in print.  The green, however, is elusive.  In some images, a yellow-green easily makes the transition from RGB to CMYK without highlights getting squashed into large pools of flat green.  In other instances, however, it is near impossible to create a yellow-green and one is only left with a green-blue (with 100% yellow and 100% cyan) that looks more like seaweed in a dank pond.

I have tried several things

1.  Returning the colorized image to grayscale (using Channel Mixer to preserve luminance) and then colorizing in CMYK.  I can't seem to colorize to green-yellow leading me to believe it doesn't dispay: but I've seen the color reproduced in publication, so I know it exists in the CMYK gamut.

2.  Pasting the RGB channels into CMY sans the K channel and then gradually adding some incremental image details borrowed from the magenta channel to the K channel (with the idea that little K value would lead to a brighter image, anyway, even if the color is off).

3. Using the Magenta channel to affect contrast.

I still get seaweed at best.

Any suggestions?  If interested, the images can be viewed and downloaded from rawlight.com/CMYKgreen.html.  Any help with this procedure would lead to helping thousands of scientists around the world!

Jerry
________________________________________________________________________

   Date: Fri, 24 Jun 2005 16:08:27 -0700
   From: Marco Ugolini
Subject: Re: Pure color green

Jerry,

I downloaded both of your images (RGB and CMYK) and right off the bat I need to ask you what profiles are supposed to be associated with each. That would give me a better idea of how to handle the color conversion, given that my system is color-managed and my monitor is calibrated and profiled.

Please let me know what profiles you are using.

By the way, even without knowing the exact profile associated with your files, I seem to be getting results that are slightly better than what you have in your CMYK file (if viewed through a "US Web Coated (SWOP) v2" profile).

Assuming that your original RGB file is in sRGB, if I convert to Euroscale Coated v2 using a Relative Colorimetric intent with Black Point Compensation (RelCol + BPC), the results are quite good: slightly cooler than the original RGB, but much yellower and brighter than what you get converting to US Web Coated (SWOP) v2 (still using RelCol + BPC).

I know, Euroscale Coated v2 is not a standard in this country, but my point is that by changing the target space for your conversion, you get different results. Also, if you convert from Euroscale Coated v2 to US Web Coated (SWOP) v2, again the colors go bluish and flat, which would indicate that there is a problem with how those bright yellow-greens are rendered within the color space of US Web Coated (SWOP) v2.

I am as puzzled as you are as to why those bright yellow-green colors should become so blue-green and dull instead: like you, I would tend to think that there should be a better match for them in the color space of the US Web Coated (SWOP) v2 profile. And things don't get much better if I convert to the US Sheetfed Coated v2 profile either.

This is all the more of a head-scratcher if you compare the gamuts of these profiles using the Grapher tool in Chromix's ColorThink 2.1.2. (Besides profiles, you can also map the colors of your color image file inside Grapher, which gives a graphic representation of where that file's colors lie within the CIELAB space.) In particular, Euroscale Coated v2 and US Sheetfed Coated v2 have an almost identical reach into the yellow-green range of colors. And yet, Euroscale Coated v2 renders those colors as much more yellow-green and vivid than does US Sheetfed Coated v2 using the same exact conversion parameters. This makes very little sense, since, if anything, US Sheetfed Coated v2 has an even BETTER reach than Euroscale Coated v2 as colors go darker.

Who can shed light on this puzzle?

--------------
Marco Ugolini
Mill Valley, CA
___________________________________________________________________________  

Date: Fri, 24 Jun 2005 19:15:57 -0500
   From: Jerry Sedgewick
Subject: Re: Pure color green

Thanks for your quick response, Marco.

On this particular image, the input profile is probably no better than sRGB, as each value in the 8-bit grayscale range is assigned the color green with a 255 G, 0 Red and 0 blue at the brightest end and 0 G, 0 R, and 0 blue at the black end.  If one were to take a 0 - 255 grayscale gradient and apply a pure green using color table in the indexed color mode, then that's about what you get. Manufacturers do not supply profiles, and the assumption is that one only needs to get appropriate color on the monitor for immediate viewing and visual satisfaction.

I have my output profile set on "U.S. Web Coated SWOP V2" because I am in the position of having to put up a fight to get calibrations/profiles from publishers.  It's an oddity in my line of work to have a scientist ask for such a thing even though many publications are changing to policies in which all authors must submit publication-ready, CMYK files.  Therefore, all I can do is set the profile generically.

I do own a Barco Monitor that is calibrated, but awaits output profiles from at least one willing publisher.  But, given the nature of this indexed-green-color-to-CMYK problem, my hope is to create some standardized approach for scientists, and let the cards fall where they may when the image is a little odd. I guess it would be a by-the-numbers approach.

I did do some further work on this image.  It appears that in the green color, whether it leans toward yellow or toward blue, the uppermost workable value lies at about 200 G.  As long as all colors lie under that (between 0 and 200 on an 8-bit scale), then everything is right with the world. Greater than that value, and the highlights get filled in with a lifeless grayish-green, and some detail blooms in the highlights.

Perhaps I should view this problem as a kind of NTSC video-world issue where the gamut is really tight.  As long as contrast can be bumped up within the limitations of dynamic range, then the image can still have enough pep to look right.

Still, I wonder if something fancy can't be done with channel blending or by ramping the K in CMYK or luminance channel in LAB.  Any thoughts would be appreciated...

Jerry Sedgewick
University of Minnesota

___________________________________________________________________________

   Date: Fri, 24 Jun 2005 22:17:30 -0400
   From: "Michael Demyan"
Subject: RE: Pure color green

Hi Sege:

In CMYK (U.S. Web coated (SWOP) v2 try these curve adjustments. My curves are with white on the left and black on the right (opposite PS defaults) Double click on the bottom shade bar below the graph to flip it of yours is the default with the black on the left. All curves are straight lines with two points.

Channel Input Output  -  Input Output
Black 0 0 82 95
Magenta 0 54 85 100
Cyan 0 25 100 80
Yellow 0 0 25 100

I got a yellow green that is close, while still preserving the shadows.
Let me know how it works for you.

Michael Demyan
Fine Photography & Digital Graphic Design
www.mikedemyan.com
___________________________________________________________________________

   Date: Sat, 25 Jun 2005 06:16:01 -0000
   From: Stephen Marsh
Subject: Re: Pure color green

Any suggestions?  If interested, the images can be viewed and downloaded
from rawlight.com/CMYKgreen.html.  Any help with this procedure would lead
to helping thousands of scientists around the world!

Jerry, some quick thoughts after playing for 5min or so.

You seem to understand the limits of CMYK and that 100CY is as pure green as you get - but this may not be the green you wish so the most saturated point may be 80c100y or whatever. As you know, you are working with a much more limited hue range and a lesser contrast range in print.

In general, no matter the final approach used, would aim for a very snappy, strong, contrasty black plate, with enhanced tonal separation that may depart a bit from the original for artistic/reproduction purposes (if acceptable, I presume you also publish the actual figures or data or whatever rather than rely only on images).

There may be less than CMY ink in the shadows of such a separation than a standard one.

A couple of quick things to try:

Starting with a great grayscale image, go to multichannel mode. dupe the black channel once, invert the tones, then dupe it again. Then add a white fourth channel. The inverted black channels will be stacked into the CY channel positions, with the blank white channel in the magenta position. The original grayscale or black channel is at the bottom. Then use mode/cmyk to change swap over from multichannel to true CMYK mode. Then assign the appropriate profile that provides a good softproof. You may need to boost the CY channels. The image will not have any CMY in the shadows though...so add a channel mixer adjustment layer: C channel = 60% of the K, M50% Y50%.

Another option may be to add any adjustment layer to the best raw conversion to CMYK that you can make. Go to layer options and uncheck the blending of the MK plates so that only CY are added. Change to multiply blending mode. One can also use blend if sliders to limit the effect to the lighter tones and not the shadows.

Another option could be to make the best conversion that you can to CMYK. Then go back to the original RGB and convert again, this time using ABSCOL rendering. Then layer this over the good CMYK in multiply mode and enter layer options to un-check the addition of the MK plates and or use blend if sliders etc.

Stephen Marsh.
___________________________________________________________________________

   Date: Sat, 25 Jun 2005 11:42:44 -0500
   From: "R. Lutz"
Subject: Re: Pure color green

Welcome to the forum Jerry. If I understood your problem I posted a possible solution in the file section of this forum:
http://groups.yahoo.com/group/colortheory/files/  Here is a summary if you don't want to use the file section.
Procedure:

1. Use the out of gamut warning to show which colors in the RGB image will not print in CMYK.
2. assign each RGB profile you have available to the RGB image and see which has the least out of gamut colors --for me it was Apple RGB.
3. Use the hue/saturation control (cmd U) to reduce the out of gamut colors while retaining something similar to the original colors.
4. Convert the RGB image to CMYK.
5. Use CMD U again and adj the sliders to get the closest color to the original RGB image.
6. Sharpen to taste
Good luck!

Dick Lutz