Understanding Color Management
Color Replication
The daily work of color-related industries such as printing, output, design and photography is color “Replication”. That is the essence of the whole industries and almost all problems are related to that important part. But how much do you know about this “replication”?
If we want to understand color management, we have to start from replication. So how are colors replicated when there is no color management?
Without a color management, colors are treated in a direct way. Take for example a CMYK color of C50M30Y0K0. When you see this color on the screen, the color is displayed after a series of treatments. Firstly the C50M30Y0K0 is converted into an RGB color through something like a logarithm from CMYK to RGB, which varies in different kinds of software and monitor. And then the RGB color is displayed directly on the screen. By so doing, the RGB color value is replicated to the screen regardless of the status or white point of the monitor. Besides, the CMYK color must be subject to the coordination among the DPI, ink, printer and paper before this color of C50M30Y0K0 may be printed out.
Just think about it: these two colors have been gone through two kinds of completely different treatments without considering the factor of vision color which means the color observed from the eyes of human beings. So how can they be the same color in observation?
From the above we may draw a simple conclusion that for an accurate replication, we must copy the vision color instead of the color value.
If we want to understand color management, we have to start from replication. So how are colors replicated when there is no color management?
Without a color management, colors are treated in a direct way. Take for example a CMYK color of C50M30Y0K0. When you see this color on the screen, the color is displayed after a series of treatments. Firstly the C50M30Y0K0 is converted into an RGB color through something like a logarithm from CMYK to RGB, which varies in different kinds of software and monitor. And then the RGB color is displayed directly on the screen. By so doing, the RGB color value is replicated to the screen regardless of the status or white point of the monitor. Besides, the CMYK color must be subject to the coordination among the DPI, ink, printer and paper before this color of C50M30Y0K0 may be printed out.
Just think about it: these two colors have been gone through two kinds of completely different treatments without considering the factor of vision color which means the color observed from the eyes of human beings. So how can they be the same color in observation?
From the above we may draw a simple conclusion that for an accurate replication, we must copy the vision color instead of the color value.
Vision Color
So how can we copy the vision color? Theoretically speaking, we must understand how to describe a vision color before we may ever replicate it. So firstly we have to introduce a color coordinate system named “Lab” (there is also LCH, XYZ, etc., but here we first take Lab as a simple example), which is different from RGB and CMYK. RGB and CMYK are based on the illuminant or ink of the device platform, while for Lab there is no such thing as a platform and it is independent from any platform.
So if you want to know the Lab coordinate value of a device color (vision color), you must have a measure device to obtain that Lab value. In other words, if you don’t have a measure device to obtain the Lab value, you can never know the Lab value of that color, and then you cannot replicate it, let alone color management. So we may draw a conclusion that without a measure device there will no real color management. Isn’t that a little bit too harsh! Actually, no. For example, if a customer provides a color, may be a Pantone color, a printed color, the color of some part of a product or a satisfactory color on the screen, how can you possibly replicate and print that color out accurately without a measure device? Of course, you may do it in the traditional way, by means of constant color correction, printing and output, until the customer is satisfied (but possibly the customer just accepts it reluctantly at last instead of being satisfied). But that is not the color management we are talking about.
We have a conclusion above that a measure device is essential to the real color management.
So how is color management with a measure device? Now let’s take a look at profiles. A profile is generated by printing out a series of patches and then measure them with a measure device. Generally a profile has two functions, firstly, converting the device color value into a Lab value, and secondly, converting the Lab value back into the device color value. Not all profiles of devices have those two functions. Some profiles may only have the first functions. It is enough for the profiles of a digital camera, for example, to just convert the device color value (RGB) into a Lab value. From the coordination of the two functions in the profiles, I think you probably start to under the keys in there. And then you should understand how color management is possible.
So if you want to know the Lab coordinate value of a device color (vision color), you must have a measure device to obtain that Lab value. In other words, if you don’t have a measure device to obtain the Lab value, you can never know the Lab value of that color, and then you cannot replicate it, let alone color management. So we may draw a conclusion that without a measure device there will no real color management. Isn’t that a little bit too harsh! Actually, no. For example, if a customer provides a color, may be a Pantone color, a printed color, the color of some part of a product or a satisfactory color on the screen, how can you possibly replicate and print that color out accurately without a measure device? Of course, you may do it in the traditional way, by means of constant color correction, printing and output, until the customer is satisfied (but possibly the customer just accepts it reluctantly at last instead of being satisfied). But that is not the color management we are talking about.
We have a conclusion above that a measure device is essential to the real color management.
So how is color management with a measure device? Now let’s take a look at profiles. A profile is generated by printing out a series of patches and then measure them with a measure device. Generally a profile has two functions, firstly, converting the device color value into a Lab value, and secondly, converting the Lab value back into the device color value. Not all profiles of devices have those two functions. Some profiles may only have the first functions. It is enough for the profiles of a digital camera, for example, to just convert the device color value (RGB) into a Lab value. From the coordination of the two functions in the profiles, I think you probably start to under the keys in there. And then you should understand how color management is possible.
The Structure of Printer or Monitor ICC Profile
Two letters “A” and “B” represent two color systems in ICC Profile.
“A” represents Device Color (RGB or CMYK) “B” represents Vision Color (Lab)
“A” represents Device Color (RGB or CMYK) “B” represents Vision Color (Lab)
The Structure of Scanner or Digital Camera ICC Profile
AtoB and BtoA are more than exchange between the vision colors and the device colors. They also have two kinds of meanings.
AtoB converts device colors to vision colors, which means it has the function or responsibility of describing device colors in the way of vision colors.
BtoA, on the other hand, converts vision colors to device colors. During that process, some source colors may lie beyond the destination device gamut, so those vision colors will be compressed into the device gamut, which results in the difference between the converted device colors and the source vision colors. That difference is already clearly reflected in the conversion process instead of remaining unknown until print it out. If we obtain compressed device color and put it into AtoB, then we can get a compressed vision color, we can even know the real difference between those two kinds of colors. Therefore the BtoA function can be said as predictive.
AtoB converts device colors to vision colors, which means it has the function or responsibility of describing device colors in the way of vision colors.
BtoA, on the other hand, converts vision colors to device colors. During that process, some source colors may lie beyond the destination device gamut, so those vision colors will be compressed into the device gamut, which results in the difference between the converted device colors and the source vision colors. That difference is already clearly reflected in the conversion process instead of remaining unknown until print it out. If we obtain compressed device color and put it into AtoB, then we can get a compressed vision color, we can even know the real difference between those two kinds of colors. Therefore the BtoA function can be said as predictive.
After understanding the two major functions of profiles, you may wonder how a profile offers color management. And that brings us back to the issue of “Replication”. Let’s take for example how a printer can print out photos in the original colors in digital photography:
The digital camera use [Device Color to Lab Color] function convert a device RGB to Lab, and then sent it out. after that, the printer profile receive the external Lab color and then, use [Lab Color to Device Color] function to get its own device color, before printing it out. We can see that the communication between those two devices require their own profiles for the replication of Lab in order to maintain Vision Color consistency.
Color Management Workflow
In the world of color management, any device, including a monitor, a printer, a printer, a scanner or a digital camera each has its own profile or profiles. Some devices may need more profiles to complete various tasks. That leads to the conclusion that any device involved in color management must have its own profile.
Here you may ask, when some or most customers do not provide images with profiles at daily work, is it impossible to handle them with color management? And wouldn’t my color management system be anything but an empty shell?
First we have to understand a key point that, upon using color management, you already have absolute control over your device outputs with confidence in every part of the work. So when you get an image file with no profile, you only have to arrange a profile for the image (firstly determining its nature), either sRGB or Adobe RGB, and then display the image on the screen. In that way you can see the profile arranged for the image arrangement and the potential results of output.
Actually most customers who are not using color management depend on two kinds of colors at work: monitor and sample colors. Let’s talk about the screen first. You only have to create several monitor profiles for different white points in advance. You don’t want too many of them. In my experience, only the ones of 5000K, 6500K, 7500K, 8000K and 9300K will be enough. In that way you can easily simulate the customer’s screen with yours. When the customer is satisfied, you just convert the simulated screen to your working screen in order to maintain the customer’s original vision requirements and also keep it under your control for printing out a satisfactory work.
Another thing customers often depend on is the sample color. It is even easier to handle it. In the previous paragraph I emphasized that color management certainly needs a measure device. So you only have to use this measure device to measure the Lab value of the sample color, before you can easily replicate and store that color and export the closest color. Those procedures are just a piece of cake for you with color management.
Now let’s go to some examples to further demonstrate how to maintain color consistency with profiles.
Here you may ask, when some or most customers do not provide images with profiles at daily work, is it impossible to handle them with color management? And wouldn’t my color management system be anything but an empty shell?
First we have to understand a key point that, upon using color management, you already have absolute control over your device outputs with confidence in every part of the work. So when you get an image file with no profile, you only have to arrange a profile for the image (firstly determining its nature), either sRGB or Adobe RGB, and then display the image on the screen. In that way you can see the profile arranged for the image arrangement and the potential results of output.
Actually most customers who are not using color management depend on two kinds of colors at work: monitor and sample colors. Let’s talk about the screen first. You only have to create several monitor profiles for different white points in advance. You don’t want too many of them. In my experience, only the ones of 5000K, 6500K, 7500K, 8000K and 9300K will be enough. In that way you can easily simulate the customer’s screen with yours. When the customer is satisfied, you just convert the simulated screen to your working screen in order to maintain the customer’s original vision requirements and also keep it under your control for printing out a satisfactory work.
Another thing customers often depend on is the sample color. It is even easier to handle it. In the previous paragraph I emphasized that color management certainly needs a measure device. So you only have to use this measure device to measure the Lab value of the sample color, before you can easily replicate and store that color and export the closest color. Those procedures are just a piece of cake for you with color management.
Now let’s go to some examples to further demonstrate how to maintain color consistency with profiles.
Production Flowchart
From the representation below we can see how a Profile maintains the Vision Color consistency of an image from photography, to Display on the screen and to output with Inkjet printer.
Proofing Flowchart
This is the most important part in the whole process. When the BtoA (user intent) of the Offset printer profile is converted to the printed device colors, some colors can be compressed. And then when Lab value are obtained with the AtoB (usually use Absolute intent) of the same profile, it is equivalent to displaying the actual situations with Lab value in advance. And then those Lab value are sent to BtoA (usually use Absolute intent) of the inkjet profile for the inkjet to display the Offset printing results.
The above representation describes the profile process of electronic proofing. You may notice that the printing profile uses a function with two directions to receive the Lab value from digital camera and then convert them into its device color. Maybe some Lab colors of the digital camera cannot be expressed (with their Lab coordinates beyond the printing gamut scope), so the printing profile uses similar colors instead. Of course the substitute colors may be a little different from the original ones, so the caputre image by the printer may not resemble the original colors in some parts. And now we would like to use an inkjet to simulate the printout of this printer; therefore that different is absolutely going to be displayed on the inkjet, so that we can observe the potential problems in advance. To capture various defects of the printer we will have to print out the substitute colors selected by the profile. After conversion back into the Lab, they are then sent to the inkjet profile, which converts the received Lab back into its own device colors and then print them out. In that way the inkjet can capture the performance of the printer.
Singular and Continuous Workflow
There are two kinds of simulation or replication process of a profile, which I call singular and continuous types respectively.
The color conversions between profiles are indicated with the two methods above: Singular Conversion and Continuous Conversion. Those two conversion methods are with totally different purposes. You may ask: Isn’t profile conversion intended for color consistency? Yes, that is the ultimate goal. But in reality they have even more complicated requirements (purposes). Let’s take for example the imitation among actors. The above profiles of X, Y and Z represent three actors respectively.
Singular Conversion:
Actor Y and Actor Z respectively imitate Actor X. Now matter how well they do that, they still demonstrate their own characters. But anyway, the point is that both Actor Y and Actor Z have to imitate Actor X as much as possible. Here is an actual example. The advertising client has designed a poster to be published in both magazine and outdoor. It certainly wants both media have the same images as the original version. So both the magazine printing and the outdoor inkjet will have to imitate the original image for the client’s satisfaction. That is a strong example of Singular Conversion.
Continuous Conversion:
Actor Y imitates the performance of Actor X while Actor Z has an even important job to imitate the interpretation of Actor X’s performance by Actor Y. How difficult it is! Needless to say, all the audience is watching Actor Z carefully to see if he can fulfill the task! That’s right. Actor Z is the leading role of this performance. His performance is difficult because he not only has to imitate Actor Y, but also has to remember that Actor Y is imitating the performance of Actor X. In other words, if Actor Y has some unique performance in interpreting the performance of Actor X, Actor Z will have to imitate that completely as well. Now let’s go back to the actual example above. The advertising client demands that you provide a proof for approval before printing the actual outdoor inkjet printing. At this time you can use a small inkjet printer to simulate the result of a actual one for the client to review. Similarly, you can use the small inkjet printer to simulate the printing result of the original document for the client to review. If you are not required to provide the proof, you can even use the monitor as the proofing machine to simulate the finished product to be printed.
In “theory”, those two kinds of processes lead to the same result, but in reality it is impossible. So why? Possible reasons:
Singular Conversion:
Actor Y and Actor Z respectively imitate Actor X. Now matter how well they do that, they still demonstrate their own characters. But anyway, the point is that both Actor Y and Actor Z have to imitate Actor X as much as possible. Here is an actual example. The advertising client has designed a poster to be published in both magazine and outdoor. It certainly wants both media have the same images as the original version. So both the magazine printing and the outdoor inkjet will have to imitate the original image for the client’s satisfaction. That is a strong example of Singular Conversion.
Continuous Conversion:
Actor Y imitates the performance of Actor X while Actor Z has an even important job to imitate the interpretation of Actor X’s performance by Actor Y. How difficult it is! Needless to say, all the audience is watching Actor Z carefully to see if he can fulfill the task! That’s right. Actor Z is the leading role of this performance. His performance is difficult because he not only has to imitate Actor Y, but also has to remember that Actor Y is imitating the performance of Actor X. In other words, if Actor Y has some unique performance in interpreting the performance of Actor X, Actor Z will have to imitate that completely as well. Now let’s go back to the actual example above. The advertising client demands that you provide a proof for approval before printing the actual outdoor inkjet printing. At this time you can use a small inkjet printer to simulate the result of a actual one for the client to review. Similarly, you can use the small inkjet printer to simulate the printing result of the original document for the client to review. If you are not required to provide the proof, you can even use the monitor as the proofing machine to simulate the finished product to be printed.
In “theory”, those two kinds of processes lead to the same result, but in reality it is impossible. So why? Possible reasons:
- Different gamut sizes of devices
- Numerical tolerance in the profile structure
- Accuracy of measuring
- Profile quality
Here, the different gamut sizes of devices have the greatest impact. Now let’s replace these two processes with different gamut sizes and see what will happen.
Singular:
The monitor and the printer have different gamuts but they both simulate the same digital camera. As a result, they both have some parts similar to the digital camera, while other parts restricted by their own gamuts. It also indicates that the monitor and the printer do not resemble each other very much.
Continuous:
Although the inkjet gamut is greater than the digital camera gamut, the inkjet is required to simulate the offset printer. However, the final exported gamut will be restricted by the offset printer gamut. And that is also the result we want. Otherwise, how can we preview the final printing result! In this picture, the inkjet completely covers the printing gamut. So theoretically the inkjet can simulate the overall printing output.
Those two methods of color replications are obviously different from each other. The first one concerns two different target profiles respectively simulating the source profile, while the second one concerns simulating the source profile and then being simulated by another profile. Generally speaking, the first method is used when an image is displayed on the screen and then printed out. And the second method is used in case of electronic proofing before mass printing.
From those two methods we can also see that complete replication is impossible as long as the destination profile is completely cover the source profile. Due to the limit of device gamut, the destination profile will use similar colors to replace those incapable of complete replication.
Although the gamut restrictions between devices have major impacts on accurate color replication, the concept of profile replication has already make the best of inter-device replication, which would have never been impossible with manual work or experience as before. Because the profile makes every possible effort for the accurate replication of colors, we can easily see the ultimate performance of devices. Therefore the user of color management may easily understand the merits and limitations of each device and thus use it better.
Now I believe you have understood how the color management system may replicate colors and what advantages and restrictions it has. You may also know about the important impacts of the device gamut size on color replication.
So do I just need to buy a device with a great gamut to have good color replication results? Not exactly. A great gamut device would be a waste if the profile quality is bad. Therefore the abilities of the profile creation software are very important. And it is wise of you to choose Qualux for that purpose. So do I have nothing more to worry about if I use Qualux together with a great gamut device to create a perfect profile?
Not yet. Color management attaches importance to the accuracy of color replication, and devices, paper, ink and screens may change along with the time. So to maintain highly accurate color replication, you must update your profile on a regular basis, at least once a month (the bottom line). Of course it also depends on your device. If you don’t update the profile for a long time, the color accuracy will be lower and lower. And eventually this profile will be similar to the old profile setting in that it loses the protection of the color management system. And consequently you will have returned to the traditional way of work without even realizing it.
Chang Fai
19 March 2008
The monitor and the printer have different gamuts but they both simulate the same digital camera. As a result, they both have some parts similar to the digital camera, while other parts restricted by their own gamuts. It also indicates that the monitor and the printer do not resemble each other very much.
Continuous:
Although the inkjet gamut is greater than the digital camera gamut, the inkjet is required to simulate the offset printer. However, the final exported gamut will be restricted by the offset printer gamut. And that is also the result we want. Otherwise, how can we preview the final printing result! In this picture, the inkjet completely covers the printing gamut. So theoretically the inkjet can simulate the overall printing output.
Those two methods of color replications are obviously different from each other. The first one concerns two different target profiles respectively simulating the source profile, while the second one concerns simulating the source profile and then being simulated by another profile. Generally speaking, the first method is used when an image is displayed on the screen and then printed out. And the second method is used in case of electronic proofing before mass printing.
From those two methods we can also see that complete replication is impossible as long as the destination profile is completely cover the source profile. Due to the limit of device gamut, the destination profile will use similar colors to replace those incapable of complete replication.
Although the gamut restrictions between devices have major impacts on accurate color replication, the concept of profile replication has already make the best of inter-device replication, which would have never been impossible with manual work or experience as before. Because the profile makes every possible effort for the accurate replication of colors, we can easily see the ultimate performance of devices. Therefore the user of color management may easily understand the merits and limitations of each device and thus use it better.
Now I believe you have understood how the color management system may replicate colors and what advantages and restrictions it has. You may also know about the important impacts of the device gamut size on color replication.
So do I just need to buy a device with a great gamut to have good color replication results? Not exactly. A great gamut device would be a waste if the profile quality is bad. Therefore the abilities of the profile creation software are very important. And it is wise of you to choose Qualux for that purpose. So do I have nothing more to worry about if I use Qualux together with a great gamut device to create a perfect profile?
Not yet. Color management attaches importance to the accuracy of color replication, and devices, paper, ink and screens may change along with the time. So to maintain highly accurate color replication, you must update your profile on a regular basis, at least once a month (the bottom line). Of course it also depends on your device. If you don’t update the profile for a long time, the color accuracy will be lower and lower. And eventually this profile will be similar to the old profile setting in that it loses the protection of the color management system. And consequently you will have returned to the traditional way of work without even realizing it.
Chang Fai
19 March 2008