Section 5.2. Profile Editing

5.1. Under the Hood

As we've already seen, color management can be a technical subject, truly understandable only by color scientists and color geeks. But a little advanced insight can be helpful for achieving true practical understanding. This chapter will review some of the earlier stuff, but with a more expert eye.

For the most part, the work that is done with color management is done under the hood. For many people, this is where it should stay. However, there are always a certain number of you who want to understand what is actually happening, and I don't blame you: the more you understand the whys and hows, the better your color, and the more likely you are to know what to do if something goes wrong.

As we've seen, in its simplest form, color management converts color between different devices to achieve predictable and repeatable color. So how is this done? As covered in previous chapters, color management consists of two steps: calibration and profiling.

5.1.1. A Closer Look at Profiles

So by now you understand that a profile is no more than the characterization of a device and how it sees, displays, or reproduces color. Profiles are broken down into various classes: display, input, output, color space conversion, abstract, named color, and device link.

We've already spent some time in this book with display, input, and output profiles, so they should be pretty familiar by now. Other profile classes that you may not be as familiar with are used for very specific purposes, and for the most part, you will not run into them as a photographer.

Profiles vary between very small (a 4K matrix-based profile of a monitor or color space) to very large (3MB for a very large table-based output profile). Each profile, no matter what the size, consists of a variety of data called tags and must conform to ICC specifications. The minimum number of tags that a profile may contain is four, which includes the name and the white point.

More complex profiles, such as a printer profile, include much more information, such as color tables for different rendering intents, both to the printer and back to the monitor for soft proofing, or for doing simulations of one device with another. If you are interested in going into the inner working of profiles further, there are numerous articles and books that go into great detail on the subject (see the Appendix for a list of our favorites).

5.1.2. Color Management Module (CMM)

As we noted in Chapter 2, the CMM is the part of the software that actually does the transformations, which involve considerable math. Both Windows and Mac operating systems include their own CMM as part of the operating system. For the most part, the two CMMs are identical, since they both were originally licensed from Linotype-Hell. CMMs are also available from other vendors, including Adobe, Logo, and X-Rite to name a few. In Photoshop, for instance, you have the option to pick your own CMM, and the default is the Adobe CMM. If you want to experiment, you can try picking a different CMM and see if there is any visual difference. Most of time, you will see either little or no difference. However, some images with different rendering intents may show greater differences when the CMM is changed.

Since the CMM does a mathematical transformation and the color tables in the ICC profile do not contain every point, the CMM has to interpolate the data. If profiles contained every point, they would be very large. How interpolation is handled is different for each vendor.

5.1.3. CMYK Proofing

The ICC-based color management system works reasonably well, but there are some limitations. One of the main limitations of ICC color management comes in preparing prepress proofs, especially when trying to match a printing press or contract proof. For prepress proofing, you are typically converting from CMYK to CMYK, and with ICC color management, the data does not convert as accurately as needed. There are a variety of digital-proofing RIPs (see "Raster Image Processors (RIPs)," later in this chapter) on the market, such as those by GMG and Dupont, that do not use ICC color management but instead use a four-dimensional table to overcome the limitations of ICC. These specialized proofing RIPs usually send a standard test target, such as the ECI 2002, which contain many patches of different CMYK combinations, to both the press and the proofing printer. Each of the four colors of the digital files is mapped directly to the CMYK of the output device. This type of transform is referred to as a four-dimensional table. It is nothing more than a large lookup table that provides a more accurate proof than can be obtained with an ICC workflow. The transfer is able to produce a more accurate proof because it converts from one profile to another by going through a profile connection space (PCS), such as CIE LAB, using a CMM.

These proofing RIPs are considerably more expensive than conventional RIPs and take a lot of work to set up.