MAX-IMIZE Your Color - Tips from our Color Scientist Max Derhak

Archived Article Excerpts - Click the monthly links for the complete ONYX Thrive Newsletter.

Thrive Newsletter - October 08

Color Management 101 - Communicating Color

In previous issues of "Max-imize Your Color" some advanced topics of Color Management were covered. This
issue covers some basic concepts related to Color Management for those who are just getting started.
When making color reproductions the first question that comes to mind is how the color of something is
communicated - and then reproduced.

In general, there are three main methods used to communicate a color:

1. The first method is to have an explicit name for each observable color. Using names for every color requires
that everyone agrees on the specific names to use for each observable color from a palette of possible colors.

Usually this means that physical examples of the colors are somehow produced with names and identifiers
associated with each color. With this method, the process of how the color is achieved is not necessary to
communicate what the color should look like. Additionally, there doesn't need to be any relationship between the
names of various colors.

Communicating a color using this method is as simple as pointing to a color and saying "I want that one."
An example of this method is using PANTONE® color charts to specify colors.
One problem with this method is that you are limited to communicating only the colors that are in the palette.
Accurately communicating colors outside the palette requires more than a simple color name.

2. The second method is to use a "recipe" that specifies how to create each color. This requires that everyone
agrees on what the ingredients are and how "mixtures" of the ingredients should be put together. This is
generally more flexible than the first method. Often colors in the first method are also defined using a recipe of
some sort.

Common examples of "recipe" based systems used to communicate colors are systems that define color using
either combinations of Red, Green, and Blue phosphor emission intensities, or combinations of Cyan, Magenta,
Yellow, and Black inks.

Only rigorous conformance to an agreed upon set of ingredients and the processes of mixing them will ensure
that a color is accurately communicated and reproduced.

This leads to some difficulties with this method. Different color reproduction technologies have different
"ingredients" and methods of putting them together. If the ingredients or processes of mixing ingredients change
in any way the resulting color will most likely be different than expected by the "recipe".
Additionally, problems come up if you want to go from one recipe system to another, or change either the
ingredient makeup or mixing method a bit.

3. The third method is to communicate color in terms of how a human observer would perceive it. In 1931 an
international standard was developed based upon color matching experiments to define a quantifiable method
of describing what a "standard observer" sees. With a definition of a "standard observer," measurement devices
can be created that measure quantities of light from an object at various wavelengths to produce a set of XYZ
numbers that define what an observer would see under a given light source.

Generally, if the measured XYZ values of two objects are the same, their color will appear the same under the
same light source for a standard observer. This is independent of the means (or recipes) used to create the
colors of the objects.

Note: In 1976 a transformation of XYZ values to Lab values was defined that allows for the description of color in
terms of Lightness, Chroma, and Hue.

Color management is made possible by combining methods two and three for communicating color. A common
standardized approach to do this is to use ICC color profiles which establish relationships between color recipes
and observable and measurable colors.

By combining ICC color profiles one can "convert" from one recipe, ingredient or mixing system to another. An
input profile associates the "recipe" for values in an image to observable colors. An output profile then converts
these observable colors into a recipe of the values needed by the output device to produce the color specified
by the image.

Note: Some colors may not be reproducible by any recipe of ingredients used by the output device, in which
case, some method of picking a "right" wrong color is used. This is referred to as "gamut mapping."
In order for color management to work, the input profile needs to match the recipe used by the image and
the output profile needs to match the recipe system provided by the output device. If the recipe, ingredient,
mixing systems of the image or output device do not match the recipe, ingredient, or mixing systems defined by
the profiles, then the color may not be communicated correctly and the reproduced color may not be correct.
Setting up profiles correctly is critical for color management to work.

Thrive Newsletter - September 08

ICC Profile Swatch Choices

When generating an ICC swatch you print a swatch, of color patches, measure the swatch, set up build options,
and then the profile build engine thinks for a while (sometimes for a long while) before saving the profile. In the
process of building a profile one often focuses on the profile build options and how to set them to get desired
results. Certainly the profile build options affect what is happening while the number crunching is going on, but
one of the biggest decisions related to how well a profile performs can be related to the choice you make when
the ICC swatch is printed.

A significant portion of building a good profile involves having a good idea of how each and every possible
ink combination relates its measured color. Establishing this relationship is known as defining a device
characterization (since we are trying to determine the device's character or behavior for each ink combination).
Having a good device characterization is essential to getting a good profile.

A large part of defining a device characterization comes from the measurement of the patches on the swatch
and the knowledge of how much ink goes into each patch. In most cases, every possible combination of ink
is not printed in the swatch. Therefore, the expected measurement of colors that are not printed needs to be
guessed at. Having more patches in a swatch can result in better device characterization since the gaps are
smaller and there is less to guess about.

(Note: In some instances, things like ink limiting and ink restrictions can make the guessing a lot more difficult).

For CMYK profiling, the Profiling Lab at ONYX generally uses an Ultra Accuracy swatch for generating ICC
profiles, since this provides the highest degree of sampling to provide the best device characterization while
building the profile. Using a swatch with fewer patches can result in a larger possibility of having a less than
perfect device characterization. In many cases there is a correlation between the time to print and measure the
swatch and the accuracy in building a profile. Larger swatches can take more time, but they can also improve
accuracy.

For printers that have more than CMYK inks (like CMYKOG or CMYKRGB ink combinations found in more
recent printers), there are two general categories of swatches that can be selected when printing the ICC
swatch. Both options have pro's and con's with no one best choice for all situations.

1. Use a CMYK based swatch.
If a CMYK image is printed to one of these printers without using ICC profiles, the CMYK ink combinations are
re-separated using hard coded ink separation to make better use of the extra available ink channels. Because
of this, it is possible to generate a CMYK profile for one of these printers, and then allow the hard coded ink reseparation
to make use of the extra inks.
Unless a swatch name identifies ink channels in its name, the swatch can be considered to be a CMYK swatch.
The CMYK swatch choices are all selectable when profiling a CMYKOG or CMYKRGB printer.

Note: When using a CMYK profile with hard coded ink separation, the ink separation is not always optimal.

Additionally, the device characterization information (used to display the soft proof of the image) does not
actually represent the device since information about only four pseudo channels is maintained.
Building a CMYK profile takes less time. Additionally, due to the nature of CMYK characterization, the profiles
can be less affected by Ink Limiting and Ink Restrictions than CMYKOG and CMYKRGB profiles.
The default swatch when profiling one of these printers is actually a CMYK based swatch.

2. Use a Swatch designed for the inks used by the printer.
For a CMYKOG printer there will be a CMYKOG swatch, and for a CMYKRGB printer there will be a CMYKRGB
swatch. In this case, different combinations of all the color channels available to the print mode are utilized by
the swatch.

CMYKOG and CMYKRGB swatches are organized differently than CMYK swatches in that they sample the
possible ink percentages much more selectively. This is because with more channels comes a LOT more
possible combinations. All possible combinations are covered, but the more likely combinations have more
patches and the less likely combinations have fewer patches. This can result in more accurate results for some
combinations and less accurate results for others.

With a CMYKOG or CMYKRGB swatch it is possible to dynamically determine the ink separation in an attempt
to define a more optimal ink separation. This option in the build settings is only selectable when a CMYKOG or
CMYKRGB swatch is used to build the profile. Because more channels are being processed building a CMYKOG or CMYKRGB profile can take significantly longer to build.

Thrive Newsletter - August 08

Color Management and Calibration

In some workflows where CMYK images are used it is desirable to set up the linearization to mimic the output of a known printing condition (like an offset press). This assumes that the inks in the printer closely match those of the printing condition. IF THIS IS THE CASE then ICC profile based color management may not be as necessary, and well defined targets for the linearization are all that are needed. Let us look at some pointers on how to go about achieving this.

(Note: Generally ICC color management is recommended to ensure color management regardless of the ink formulation. Additionally, ICC profiles are generally required for Spot Color Replacement, as well as RGB workflows). When profiling a mode, the next step after printing and reading the calibration swatch is the "Target Densities" step. This step is used to determine the target densities to aim for when building the linearization and whenever recalibration is performed. In the Linearization step one can view and modify the target densities before building the linearization.

The "Target Densities" step has two modes of operation. The simplest and most often used method is to generate "Basic Density Curves." This generates basic target densities that are a fairly linear progression from paper to full coverage for each output primary. With this method there is no implied relationship between the ink channels. Combinations of inks have no predetermined result. The "Advanced Grayscale" mode of operation allows one to generate target densities that provide for an initial approximation at getting a balanced gray when using CMY combinations. This is accomplished by printing a swatch and measuring it to set up the target densities. The "Setup..." for printing the Advanced Grayscale swatch has a "Black Setup" tab that allows one to determine the approach to balancing the grayscale.

If Multi-Point Gray Balance is used then three N-factors are calculated for each CMY channel to balance the grayscale at 25, 50, and 75%. N-factors are not calculated for other points and interpolation is used to set target densities for other points. As such, the gray balance may need additional manual adjustment in the Linearization step.

The Desired Gray Balance setup is used to determine what combinations of CMY should be gray. Clicking "Equal CMY" causes equal amounts of CMY to appear gray. Clicking "Offset Press" causes CMY percentages to be set to appear gray in a manner consistent with offset press printing. Additionally the percentages for magenta and yellow can be clicked on and manually changed to allow any desired combination to be configured for gray appearance.

Turning off the Multi-Point Gray Balance results in a single N-factor being calculated for each CMY channel to balance the gray at a midtone. The Midtone Black Setup can be used to determine where to place the midtone. The CMY Gray Setup can be used in either case to determine the interpretation of gray. Colorimetric gray with a specified target using CIELAB's a* and b* values can be selected, or the CMY gray balance can be set to match the appearance of using the black channel.

Using target densities to define the gray balance is generally a time-consuming process that requires printing out a gradient and tweaking the target densities as needed. The Advanced Grayscale target density generation step provides an initial stab at setting up a balance linearization though further refinement may still be needed.

Ink limiting and ICC profile generation can still be performed in addition to using a gray balanced linearization. This facilitates mixed workflows that support both ICC profile and non ICC profile settings in Quick Sets.

Thrive Newsletter - July 08

Using Achromatic UCR

The ONYX ICC profile build options provide the ability to use UCR or "Under Color Removal" for the purposes of
black generation for achromatic (or gray) portions of images.If the UCR option is turned off (the default) the Start Black option defines where "black" ink is introduced relative to CMY inks in the color Mix. The above figure shows this relationship.

Note: This is the traditional method of performing black generation.

This results in potentially smoother highlights as the light areas are made up of CMY dots. However, the gray
balance can be more unstable if the CMY target ink densities are not properly color balanced, and the use of
CMY inks to render grays can also be less color constant (IE: they can change appearance to different observers
or in different lighting conditions).

If the UCR option is turned on, the Start Black option defines where CMY inks are introduced relative to "black"
ink in the color mix (IE: the CMY inks are removed from the highlights). The above figure shows this relationship.
This results in potentially coarser highlights as lighter areas are only made up of black dots. However, if light
"black" ink(s) or gray ink(s) are available in the print mode this is usually not a problem. With only "black" ink being
used to render the highlights the gray balance will be much more stable and color constant.
TIP: If you are profiling to print black & white photography on a printer that has light blacks, consider using the
UCR option get better grayscales.

Thrive Newsletter - June 08

Difference between "Device Gamut" and "ICC Profile Output Gamut"

The gamut of a device is the range of colors that can be achieved using all possible combinations of device
values (ink percentages for each output channel). The gamut of an ICC output profile is the range of colors that
can be achieved using the profile. Different profiles can have different output gamuts depending upon how they
are built.

A profile's output gamut size and shape can be affected by:
- Number of grid points
- Color Separation
- Black Generation
- Accuracy of Interpolation
- Total Ink Usage

The accompanying image shows the difference between a device gamut and
an ICC output profile gamut. The difference is mostly a result of the choice of
chromatic start black and accuracy of interpolation.
Note: Some profile gamut viewers ONLY display device gamut rather than the
profile's output gamut.

Tip: Rendered gradients of out of gamut colors are affected by the shape of
the profile's output gamut. If you have problems you may consider lowering the
chromatic start black. However, in some cases this may also introduce black
dots that could make the image appear more grainy.

Thrive Newsletter - May 08

Ink Restrictions vs. Ink Limits

In the ONYX Version 6.5 Media Manager application it was important to not use low Ink Limits because low Ink Limits would cause the ICC profile to produce unpredictable results. To get higher Ink Limits it was required that you set Ink Restrictions lower.

Since Ink Restrictions are much less selective than Ink Limits the available color gamut with lower Ink Restrictions is generally
smaller than when using Ink Limits. A lot of work was put into the ONYX Version 7 ICC Profile Generation engine to allow lower Ink Limits (around 1.5 to 2.0) to be used with the Advanced Ink Limit dialog with CMYK (plus light ink) printers. The Advanced Ink Limit settings allow for more selective reduction of available ink use and a larger color gamut.

Recommendation: If you set the Ink Limits low you will most likely want to set the Black Ink Compensation higher to ensure that transitions going from "two color" combinations to "two color plus black" ink combinations do not turn grey. With really low ink limits the Black Ink Compensation may need to be set even up to 5.0. Re-print the ink limit swatch after changing ink limit settings to ensure that you get desirable results.