akaracquel
New member
Gosh - it's very quiet in here. Hello quiet space 
CYMK Ouch
- Thread starter akaracquel
- Start date
DrStrik9
Scene Destroyer
Remember what color gamut is?
<Stretching arms out as far as possible> "This is how many colors we can see in RGB."
<Holding fingers about 6 cm apart> "This is how many colors we can see in CMYK."
Now sing, "Y-M-C-K" - to the Village People tune ...
In fact, in magazines, and anything printed on paper using the CMYK SWOP standard, everything looks great ONLY because it's not being compared to anything else. The mind's eye compensates for the extremely limited CMYK color gamut big-time, within a viewing environment. (The human visual cortex is AMAZING.) To see the truth of this, try finding the brightest ORANGE (or violet) color on a printed CMYK page and holding almost any real-life orange (or violet) object near the page, and the whole page instantly DIES.
This is because CMYK is the best that can be done consistently with a SUBTRACTIVE color model. By comparison, RGB is a TRANSMISSIVE color model, a wholly different (and far more beautiful) animal. (Subtractive means the only light you see is what is NOT subtracted from the original, in reflected light (think tempura finger-paints from nursery school). Transmissive means the light is being transmitted to your eye, as a light source, such as a computer monitor, or real life.)
The best you can do that I've found to "beat" this reality is to open the CMYK file in Photoshop, select the various areas that are offensively dull, and reduce the non-essential colors. For instance, if the color is violet, as in your example, try using levels or curves to reduce or remove the other colors (yellow mostly, and MAYBE black). In many cases, this can help, but typically you need to leave a "trace" (at least 3-5%) of the offending colors (depending on context), or the result looks "faked" when printed. But nothing about YMCK can ever measure up to RGB. It's just physics.
<Stretching arms out as far as possible> "This is how many colors we can see in RGB."
<Holding fingers about 6 cm apart> "This is how many colors we can see in CMYK."
Now sing, "Y-M-C-K" - to the Village People tune ...
In fact, in magazines, and anything printed on paper using the CMYK SWOP standard, everything looks great ONLY because it's not being compared to anything else. The mind's eye compensates for the extremely limited CMYK color gamut big-time, within a viewing environment. (The human visual cortex is AMAZING.) To see the truth of this, try finding the brightest ORANGE (or violet) color on a printed CMYK page and holding almost any real-life orange (or violet) object near the page, and the whole page instantly DIES.
This is because CMYK is the best that can be done consistently with a SUBTRACTIVE color model. By comparison, RGB is a TRANSMISSIVE color model, a wholly different (and far more beautiful) animal. (Subtractive means the only light you see is what is NOT subtracted from the original, in reflected light (think tempura finger-paints from nursery school). Transmissive means the light is being transmitted to your eye, as a light source, such as a computer monitor, or real life.)
The best you can do that I've found to "beat" this reality is to open the CMYK file in Photoshop, select the various areas that are offensively dull, and reduce the non-essential colors. For instance, if the color is violet, as in your example, try using levels or curves to reduce or remove the other colors (yellow mostly, and MAYBE black). In many cases, this can help, but typically you need to leave a "trace" (at least 3-5%) of the offending colors (depending on context), or the result looks "faked" when printed. But nothing about YMCK can ever measure up to RGB. It's just physics.
akaracquel
New member
No qualms about singing the C with a rude word when i'm doing the Village People dance
Tweaked it as best as i could with PS gymnastics. Not sure if it's worth reworking in LW to make adjustments at this stage because the proofs worked out well. Sensing it will help to opt for a gloss finish & stay away from matt cello glaze for this round, doing test samples to see how it goes.
Had difficulty understanding from what i read online, if there is a method available that can help preview or give a heads-up on this issue while working within LW now?
Saw a post online about SG_CCTools - haven't been able to successfully install it so I don't know if it can help any further with this, compared to using photoshop as the primary guide for reviewing & choosing RGB colours within the cmyk phuket bucket.
Tweaked it as best as i could with PS gymnastics. Not sure if it's worth reworking in LW to make adjustments at this stage because the proofs worked out well. Sensing it will help to opt for a gloss finish & stay away from matt cello glaze for this round, doing test samples to see how it goes.
Had difficulty understanding from what i read online, if there is a method available that can help preview or give a heads-up on this issue while working within LW now?
Saw a post online about SG_CCTools - haven't been able to successfully install it so I don't know if it can help any further with this, compared to using photoshop as the primary guide for reviewing & choosing RGB colours within the cmyk phuket bucket.
JonW
New member
There is some very good information here.
http://www.cambridgeincolour.com/tutorials/color-space-conversion.htm
http://www.cambridgeincolour.com/tutorials/color-space-conversion.htm
DrStrik9
Scene Destroyer
In my opinion, the best "color calibrators" are right there on your face, above your nose.
I guess while working in LW, just try to remember which colors are the worst to reproduce BRIGHTLY in YMCK, and tone them down in RGB. Early on, I did tests of this, and got a feel for the YMCK color gamut in general, and over time got better at aiming at it in LW.Of course, the other thing to do would be to quickly render out a small (low-res) frame, open it in Photoshop, do your conversion to CMYK and take a look, then go back and make adjustments in LW (RGB).
JonW
New member
Here is a good online test to see how good your monitor, calibration & eyes are performing.
http://www.xrite.com/custom_page.aspx?PageID=77&Lang=en
http://www.xrite.com/custom_page.aspx?PageID=77&Lang=en
akaracquel
New member
Enjoyed the Cambridge in Colour site, the colour test as well Feeling more confident about my monitor & the peepers on my noggin now, thanks. :thumbsup:
Feeling more confident about my monitor & the peepers on my noggin now, thanks. :thumbsup:
funk
Professional Newbie
Thanks for this nice test! This gives me more confidence about my monitor (and eyes). I scored a perfect 0, which surprised me
Last edited:
JonW
New member
One should really calibrate their monitor/s. The test is really about checking your eyes. Having a calibrated monitor is going to make it easier to do the test.
Before I used hardware calibration I could only get about 3. Years ago I actually had my eyes tested & they are a touch weak in the yellow/green spectrum.
I had to get hardware calibration because it was impossible to set my Dell 3011 to any level of usable calibration. It was very frustrating.
But now a print from the laser printer looks within reason what's on the screen.
I run the screens at 5800k 90 cd/m2 which looks better compared to paper. 6500k is a bit too cold & 120 cd/m2 is too bright.
Also all the greys from black through to white are very good, a smooth tone with no banding of any colours.
Before I used hardware calibration I could only get about 3. Years ago I actually had my eyes tested & they are a touch weak in the yellow/green spectrum.
I had to get hardware calibration because it was impossible to set my Dell 3011 to any level of usable calibration. It was very frustrating.
But now a print from the laser printer looks within reason what's on the screen.
I run the screens at 5800k 90 cd/m2 which looks better compared to paper. 6500k is a bit too cold & 120 cd/m2 is too bright.
Also all the greys from black through to white are very good, a smooth tone with no banding of any colours.
Last edited:
gerardstrada
New member
A well calibrated monitor implies that this is adjusted to a standard. Of course, higher the standard (and the system), better the color reproduction. As general rule, if the specs are not well adjusted according to the standard, then it could be an issue either in the predefined non-linear response, chromaticities or viewing conditions. If in spite of that, you don't get a good color reproduction, then the issue is probably in the color workflow. As general rule also, if you don't have a system for profiling and calibrating, then install and use (in your OS) the color profile that comes with your monitor drivers, this will be always better (sometimes surprisingly much better) than the generic profile assigned by default by the OS.
Let's remember that RGB and CMYK are color models and in the same way there are many RGB color spaces, there are also many CMYK color spaces with different gamut ranges. Gamuts are three-dimentional and even when USWebCoatedSWOP2 is smaller than sRGB, small portions of greens, cyans and a bit of yellows are not contained by sRGB.
Modern CMYK color spaces (mostly those working for multichannel printers or Giclee printing) may have wider gamuts, not all real colors, but a sRGB ouput will surely not be enough.
A comparison of these profiles can be visualized more easily in a gamut 3D viewer. Don't know many free apps that can do this, but on Mac, the Colorsync control panel can be useful. I remember there was a Color Control Panel Applet for Windows XP but don't know if it works with Win7. There's also PerfX Gamut Viewer for Win and Mac which can plot 4 profiles at the same time. The trial version of Gamutvision might help too.
After checking, if the difference of reproducible colors between RGB and CMYK is not much, a color conversion at the end of the process (in PS, LR or similar), may be enough. But if the gamut difference is too high in our final colors, we'll want to know how things will look like from the moment the image is generated. This phase is when is more easy and flexible adjust the color appearance and there's where a CMYK preview is more convenient.
Press Release of LW 10.1 said that it supports color space correction for CMYK data and LW documentation addendum (What's new in v10.1) says that linear color space supports ICC/ICM monitor profiles and after that, between parenthesis, it states that it supports, among other color models, CMYK data. It's confusing because one might think that this data could be got through ICC/ICM profiles - which is the standard profile format that has proven to work for this type of conversions in print industry - but pitifully this is not the case. Not for previewing or converting from/to CMYK ICC profiles (DepthX supports JPGs with CMYK (YCCK) data, btw). It should be referring to the LUT color tables, but without a Color Management (CM) system to generate and handle color models conversions and color space conversions through LUTs appropriately, which is prohibitive for the common CG artist/pro or small/medium studio, this is not a viable option. Good LUTs generators are not cheap neither (LUT generation services can cost between $100 to $5000) and a 3D LUT alone won't work as expected for this, since the ASC/CDL LUT standards have not defined specifications about color spaces, viewing conditions, nor gamut mapping; then without those expensive LUT-based CM systems, able to handle these aspects and others (and if they even can handle 4D LUTs) - there's no way to get a predictable and consistent color appearance. Even several LUT-based CM modules have some serious issues with this type of color conversions (mostly for device-dependent representations) - we are talking about a 4 channel color model here - guess that's why ICC/ICM profiles are used instead (besides, ICC profiles are either cheap or free). Other situation is that for print work (and for film work too) we need concatenate color conversions, seeing that a single one (and a single type of LUT) can not handle the right color appearance appropriately in these cases. But 3D LUTs that concatenate color space conversions won't work well without a CM module. Also, concatenating single color tables is not currently possible in the CS panel and it seems the useful db&w Colour Space node doesn't load ICC/ICM profiles like in the CS panel, and guess that a CMYK ICC profile won't work there too since these are output profiles, not display profiles. LW would need a strongest ICC/ICM support for solving these color conversions natively in a cheaper and easier way so that the common user or small/medium studio can have available these kinds of previews.
SG_CCTools can handle CMYK data, but the available version presents artifacts in this type of conversions. I use here an unreleased version that doesn't present artifacts (and works in LW11), but CMYK conversions are not so straightforward since the color and vector inputs/outputs only allow 3 channels and CMYK data needs 4 input channels, so there's a corrupted output in the SG_CCNode. I've been able to solve this with an Abstract profile, which basically performs the conversion internally in a single profile and outputs the correct colors. Sebastian Goersch (the developer) is working in a new version in his spare time (that promises to be better than the one I currently use), but due to the improvements of the CMM, it's like re-doing the whole thing from scratch again, and it seems this gonna take time. If you guys and girls are interested, please let him know that a new version will be very welcome :thumbsup:. His email (or a donation link) is in the SG_CCTools web page.
The results we can get in this way compared with Photoshop are these:
Now, without a preview possibility within LW, there's still a way to get a rough approach. I'd recommend to use the SG_CCPicker (free), or Jovian Color Picker (commercial) to load a screen capture of a CMYK pantone/color wheel within LW to guide your color choices in RGB space. The pantone should contain the most saturated colors that the CMYK model is able to reproduce, so that you don't surpass them. It's not ideal since color/hues relationships won't be the same, but it's better than nothing.
To install SG_CC Picker, just place it in your plugins folder, add it manually in LW and select it in the General Preferences as your color picker. A screen capture of a color wheel or pantone (up to eight) can be saved as images of 128x384 pixels and load them in the Config Panel of the picker. SG_CCPicker is able also to pick colors from any other image or app running on the screen.
The installation of the SG_CCNode and the SG_CCFilter is indeed more complex due to the color profiles list. Besides, the released version doesn't work in LW11, but if someone wants to try in LW9.6, I would recommend skip the python script for the installation since without a previous version of Python, the script won't work. Then it's easier install them by hand. Create a new folder in your plugins folder (SG_CCTools or something like that) and unzip the plugins there. In the same folder, create a .txt file called colorprofiles.txt. Identify the route in your system of each color profile you want to use in the filter or the node and copy each route there. The colorprofiles.txt would be something like this:
C:\Program Files (x86)\Common Files\Adobe\Color\Profiles\AdobeRGB1998.icc
C:\Program Files (x86)\Common Files\Adobe\Color\Profiles\sRGB Color Space Profile.icm
C:\Program Files (x86)\Common Files\Adobe\Color\Profiles\ProPhoto.icm
C:\LogProfiles\MyMonitor.icc
...and so on. Then within LW, install the plugins as usual.
In case someone wants to use SG_CCTools for preview more accurate colors in linear workflow, you can create the linear profiles with Photoshop as is described at the end of this post.
Sorry for the missing images in the linked thread but the server that stores my images has been hacked - until solving that, if the linearization method is not understandable without the images, I can re-post the method here with new images.
Gerardo
Let's remember that RGB and CMYK are color models and in the same way there are many RGB color spaces, there are also many CMYK color spaces with different gamut ranges. Gamuts are three-dimentional and even when USWebCoatedSWOP2 is smaller than sRGB, small portions of greens, cyans and a bit of yellows are not contained by sRGB.
Modern CMYK color spaces (mostly those working for multichannel printers or Giclee printing) may have wider gamuts, not all real colors, but a sRGB ouput will surely not be enough.
A comparison of these profiles can be visualized more easily in a gamut 3D viewer. Don't know many free apps that can do this, but on Mac, the Colorsync control panel can be useful. I remember there was a Color Control Panel Applet for Windows XP but don't know if it works with Win7. There's also PerfX Gamut Viewer for Win and Mac which can plot 4 profiles at the same time. The trial version of Gamutvision might help too.
After checking, if the difference of reproducible colors between RGB and CMYK is not much, a color conversion at the end of the process (in PS, LR or similar), may be enough. But if the gamut difference is too high in our final colors, we'll want to know how things will look like from the moment the image is generated. This phase is when is more easy and flexible adjust the color appearance and there's where a CMYK preview is more convenient.
Press Release of LW 10.1 said that it supports color space correction for CMYK data and LW documentation addendum (What's new in v10.1) says that linear color space supports ICC/ICM monitor profiles and after that, between parenthesis, it states that it supports, among other color models, CMYK data. It's confusing because one might think that this data could be got through ICC/ICM profiles - which is the standard profile format that has proven to work for this type of conversions in print industry - but pitifully this is not the case. Not for previewing or converting from/to CMYK ICC profiles (DepthX supports JPGs with CMYK (YCCK) data, btw). It should be referring to the LUT color tables, but without a Color Management (CM) system to generate and handle color models conversions and color space conversions through LUTs appropriately, which is prohibitive for the common CG artist/pro or small/medium studio, this is not a viable option. Good LUTs generators are not cheap neither (LUT generation services can cost between $100 to $5000) and a 3D LUT alone won't work as expected for this, since the ASC/CDL LUT standards have not defined specifications about color spaces, viewing conditions, nor gamut mapping; then without those expensive LUT-based CM systems, able to handle these aspects and others (and if they even can handle 4D LUTs) - there's no way to get a predictable and consistent color appearance. Even several LUT-based CM modules have some serious issues with this type of color conversions (mostly for device-dependent representations) - we are talking about a 4 channel color model here - guess that's why ICC/ICM profiles are used instead (besides, ICC profiles are either cheap or free). Other situation is that for print work (and for film work too) we need concatenate color conversions, seeing that a single one (and a single type of LUT) can not handle the right color appearance appropriately in these cases. But 3D LUTs that concatenate color space conversions won't work well without a CM module. Also, concatenating single color tables is not currently possible in the CS panel and it seems the useful db&w Colour Space node doesn't load ICC/ICM profiles like in the CS panel, and guess that a CMYK ICC profile won't work there too since these are output profiles, not display profiles. LW would need a strongest ICC/ICM support for solving these color conversions natively in a cheaper and easier way so that the common user or small/medium studio can have available these kinds of previews.
SG_CCTools can handle CMYK data, but the available version presents artifacts in this type of conversions. I use here an unreleased version that doesn't present artifacts (and works in LW11), but CMYK conversions are not so straightforward since the color and vector inputs/outputs only allow 3 channels and CMYK data needs 4 input channels, so there's a corrupted output in the SG_CCNode. I've been able to solve this with an Abstract profile, which basically performs the conversion internally in a single profile and outputs the correct colors. Sebastian Goersch (the developer) is working in a new version in his spare time (that promises to be better than the one I currently use), but due to the improvements of the CMM, it's like re-doing the whole thing from scratch again, and it seems this gonna take time. If you guys and girls are interested, please let him know that a new version will be very welcome :thumbsup:. His email (or a donation link) is in the SG_CCTools web page.
The results we can get in this way compared with Photoshop are these:
I haven't seen this kind of CMYK previews in no other 3D package before
Now, without a preview possibility within LW, there's still a way to get a rough approach. I'd recommend to use the SG_CCPicker (free), or Jovian Color Picker (commercial) to load a screen capture of a CMYK pantone/color wheel within LW to guide your color choices in RGB space. The pantone should contain the most saturated colors that the CMYK model is able to reproduce, so that you don't surpass them. It's not ideal since color/hues relationships won't be the same, but it's better than nothing.
To install SG_CC Picker, just place it in your plugins folder, add it manually in LW and select it in the General Preferences as your color picker. A screen capture of a color wheel or pantone (up to eight) can be saved as images of 128x384 pixels and load them in the Config Panel of the picker. SG_CCPicker is able also to pick colors from any other image or app running on the screen.
The installation of the SG_CCNode and the SG_CCFilter is indeed more complex due to the color profiles list. Besides, the released version doesn't work in LW11, but if someone wants to try in LW9.6, I would recommend skip the python script for the installation since without a previous version of Python, the script won't work. Then it's easier install them by hand. Create a new folder in your plugins folder (SG_CCTools or something like that) and unzip the plugins there. In the same folder, create a .txt file called colorprofiles.txt. Identify the route in your system of each color profile you want to use in the filter or the node and copy each route there. The colorprofiles.txt would be something like this:
C:\Program Files (x86)\Common Files\Adobe\Color\Profiles\AdobeRGB1998.icc
C:\Program Files (x86)\Common Files\Adobe\Color\Profiles\sRGB Color Space Profile.icm
C:\Program Files (x86)\Common Files\Adobe\Color\Profiles\ProPhoto.icm
C:\LogProfiles\MyMonitor.icc
...and so on. Then within LW, install the plugins as usual.
In case someone wants to use SG_CCTools for preview more accurate colors in linear workflow, you can create the linear profiles with Photoshop as is described at the end of this post.
Sorry for the missing images in the linked thread but the server that stores my images has been hacked - until solving that, if the linearization method is not understandable without the images, I can re-post the method here with new images.
Gerardo
gerardstrada
New member
Well, LW development team might contact him to help in this subject, I guess.
At the moment, I'm still using a concatenation of 2 color conversions with SG_CCNode/Filter, but I think I could pack all concatenations in a single profile.
This is another test:
It seems the PS version is a bit more contrasted. It's notorious also in the black borders of the package design of the previous sample. A slight adjustment of contrast/sat can match better the PS results:
But I'm sure is possible get exactly the same contrast with the profiles conversion only. The slight mismatch in contrast is due to the K channel in the device-link profile that is not entirely black. Have to figure out how to solve that, but the current technique has been very useful so far.
Gerardo
At the moment, I'm still using a concatenation of 2 color conversions with SG_CCNode/Filter, but I think I could pack all concatenations in a single profile.
This is another test:
It seems the PS version is a bit more contrasted. It's notorious also in the black borders of the package design of the previous sample. A slight adjustment of contrast/sat can match better the PS results:
But I'm sure is possible get exactly the same contrast with the profiles conversion only. The slight mismatch in contrast is due to the K channel in the device-link profile that is not entirely black. Have to figure out how to solve that, but the current technique has been very useful so far.
Gerardo
gerardstrada
New member
A couple of good news. Just have noticed the available version of SG_CCTools still works in LW11!!! I've solved also the mismatch in contrast with Photoshop in my profiles concatenation.
Take a look:
exactly like Photoshop!
Released version of SG_CCTools is compatible with ICC v2.x only (but many profiles are still v2.x) and color conversions work well in image filter (pre-/post-processing) and in DP Image Filter Node Editor. However in DP Pixel Filter Node Editor or Surface Node Editor, it presents artifacts when using more than one Thread (MultiThreading).
Then, for people who want to try a preview from CMYK (USWebCoatedSWOP v2 in this case) to sRGB, download the attached profiles and set them up in this way:
First SG_CCNode / SG_CCFilter instance:
Input Profile: Linear sRGB.icm
Output Profile: Lab.icc
Rendering Intent: Perceptual
Second SG_CCNode / SG_CCFilter instance:
Input Profile: CMYKpreview.icc
Output Profile: sRGB.icc
Rendering Intent: Perceptual
Note: Lab.icc can be found here. sRGB can be found here, or here, or in your Abobe profiles folder or in your OS profiles folder. The linear version of this profile (or any output profile) can be created in Photoshop as explained here.
Have fun!
Gerardo
P.D: If you find it useful, please don't forget let know to Sebastian Goersch how useful are the SG_CCTools - his email is here.
Take a look:
exactly like Photoshop!
Released version of SG_CCTools is compatible with ICC v2.x only (but many profiles are still v2.x) and color conversions work well in image filter (pre-/post-processing) and in DP Image Filter Node Editor. However in DP Pixel Filter Node Editor or Surface Node Editor, it presents artifacts when using more than one Thread (MultiThreading).
Then, for people who want to try a preview from CMYK (USWebCoatedSWOP v2 in this case) to sRGB, download the attached profiles and set them up in this way:
First SG_CCNode / SG_CCFilter instance:
Input Profile: Linear sRGB.icm
Output Profile: Lab.icc
Rendering Intent: Perceptual
Second SG_CCNode / SG_CCFilter instance:
Input Profile: CMYKpreview.icc
Output Profile: sRGB.icc
Rendering Intent: Perceptual
Note: Lab.icc can be found here. sRGB can be found here, or here, or in your Abobe profiles folder or in your OS profiles folder. The linear version of this profile (or any output profile) can be created in Photoshop as explained here.
Have fun!
Gerardo
P.D: If you find it useful, please don't forget let know to Sebastian Goersch how useful are the SG_CCTools - his email is here.