Color Settings with More Options

When you click the More Options button in the Color Settings dialog box, you gain access to the Conversion Options and Advanced Controls, as well as to a wider range of profiles (we've discussed that earlier in the chapter). The Conversion Options can be useful in typical workflows, but the Advanced Controls are a grab-bag of options that may be useful to a very small number of serious players, and are "hurt-me" buttons for almost everyone else.

Conversion Options

The Conversion Options section of the dialog box lets you control useful things like Photoshop's default rendering intent and color management engine (CMM)things you probably won't need to change very often, but might want to occasionally (see Figure 5-26).

Figure 5-26. Conversion Options

Engine

The Engine popup menu lets you select the CMM that Photoshop uses for all its color space calculations. The options that appear on the menu depend on which CMMs are installed on your system. Unless you have really pressing reasons to use a different CMM, we recommend sticking with the Adobe (ACE) engine. Bruce has found bugs in all the other CMMs he's tried (though they're pretty dang obscure), and has yet to find any in ACE. Of course, no complicated software is entirely without bugs.... When the engines work correctly, there is only a tiny change in pixel values with the different engines; except for the bugs, we've never noticed a visual difference.

Mac users will notice separate entries for the Apple CMM and Apple ColorSync. Apple CMM means that the Apple CMM will always be used. Apple ColorSync uses whatever CMM is set in the ColorSync System Preferences (which have disappeared in Tiger).

Intent

The Intent popup menu is significantly more useful for the average user. Intent lets you choose the default rendering intent that Photoshop uses in any of the following color space conversions:

• Converting documents on opening

• Converting documents by choosing a different mode from the Mode menu (like separating RGB to CMYK)

• Calculating the numbers that appear on the Info palette for color modes other than the one the document is in

• Calculating the numbers that appear in the Color Picker for color modes other than the one the document is in

Every other feature in Photoshop that lets you convert colors from one profile's space to another has its own rendering intent controls. Photoshop's default rendering intent is relative colorimetric with Black Point Compensation. These defaults are always the subject of some pretty heated debate. We offer three observations:

• While some users may prefer perceptual rendering as the default choice, rendering intent is best treated as an image-specific and conversion-specific decision. Relative colorimetric renderings may actually do a better job than perceptual rendering when your images don't have a lot of significant out-of-gamut colors.

• Photoshop's color conversions have always been relative colorimetric in the past.

• It's just a default. Don't get your knickers in a twist; you can always override it to suit the needs of the image at hand.

Since Photoshop CS2 makes it very easy to both preview and apply different rendering intents for each image, the rendering intent setting here is more a matter of convenience than of necessity.

Use Black Point Compensation

Use Black Point Compensation, when turned on, maps the black of the source profile to the black of the target profile, ensuring that the entire dynamic range of the output device is used. In many cases you'll find no difference whether it's turned on or off, because it depends on the contents of the particular profiles involved. See the sidebar "Black Is Black (or Is It?)" for a detailed look at the Black Point Compensation feature.

While in earlier versions of Photoshop we offered a complex set of rules about when to turn this on and off, we now recommend that you just leave Use Black Point Compensation turned on at all times.

Use Dither (8-bit/channel images)

The Use Dither feature is somewhat esoteric. All color space conversions in Photoshop are performed in a high-bit space. When Use Dither is turned on, Photoshop adds a small amount of noise when the 8-bit channels are converted into the high-bit space. This makes banding or posterization much less likely to occur (that's a Good Thing). But if your final output is JPEG, this tiny dithering is likely to produce a larger file size (because it introduces more discrete colors into the image), and if you're using Photoshop for scientific work, where you need to perform quantitative analysis on colors, you should turn this off, as it will introduce noise in your data. Otherwise, leave it on.

Advanced Controls

The Advanced Controls are aptly named. We recommend leaving them alone unless you have a pressing need to do otherwise, but in the interest of full disclosure, here's what they do.

Desaturate Monitor Colors By

The Desaturate Monitor Colors feature attempts to solve a problem with large-gamut working spaces: Rendering to the monitor is always relative colorimetric, so colors in the working space that lie outside the monitor gamut get clipped to the nearest equivalent the monitor can display. You can think of desaturating the monitor as the poor man's perceptual rendering. Unless you're working with a very large space like Kodak's ProPhoto RGB, don't even think about messing with this. Even then, we don't use itwe find that the problem is much less severe than theory would lead one to expect.

Black Is Black (or Is It?)

We usually think of black as being "just black," but of course black on different devices appears differently (solid black on newsprint is much grayer than solid black on glossy sheetfed stock, for instance). Photoshop's Black Point Compensation forces us to think about this fact. The information here is fairly complex, but the basic principle is simple. When you transform from one color space to another, there are two ways of transforming the black point: absolute and relative.

Transformations involve first mapping the source gamut to the reference color space (also known as the Profile Connection Space, or PCS), which in most cases is Lab, and then mapping the Lab values to the destination space. In a relative black-point transformation, the source black is mapped to a L* value of 0 in the PCS, but in an absolute black-point transformation, it's mapped to the actual L* value that the source device can produce, which is usually substantially higher than zero. (A zero L* value represents the total absence of any reflected light, which is blacker than anything other than a black hole can reproduce.) The ICC profiles themselves specify whether the transform should be absolute or relative.

This can lead to undesirable results. For example, Radius ColorMatch RGB profiles map RGB 0,0,0 to L*a*b* 3,0,0 in the PCS. A CMYK profile that uses absolute black transformation may map to a black value in the PCS of L*a*b* 7,0,0. If you convert an RGB image to CMYK using this pair of profiles, your shadow detail will get clobbered because the first few levels in the RGB document will convert to L* values in the PCS between 3 and 7. Since these are all darker than the output device can produce, they'll be clipped to black, and your shadow detail goes bye-bye.

If the same RGB profile is used with a CMYK profile that maps device black to L*a*b* 2,0,0, you'll get very different, equally undesirable results. The RGB black will convert to L*a*b* 3,0,0, which is lighter than the black the output device can produce. The resulting image will appear slightly washed out because it contains no true blacks.

If you want a transform that uses the entire dynamic range of the output device, you need a relative black transform both from source to PCS and from PCS to output. Photoshop's Black Point Compensation forces this to happen, no matter what the profiles say. It works by estimating the black point for the source and the target.

If they're the same, as they would be if both profiles use relative black encoding, the feature does nothing. But if the black levels are different, it adds an extra processing step: After the source color is converted into the PCS, Black Point Compensation adjusts the PCS to map the source profile's black to the destination profile's black via a straightforward linear transformation of the L* values in the PCS. This ensures that the entire dynamic range of the source is mapped into the entire dynamic range of the target, without shadow clipping or washed-out blacks. Again, we now turn this on and leave it on.

For those brave (or foolhardy) souls who want to experiment with it, a setting in the 12 to 15 percent range seems somewhat useful for Kodak ProPhoto RGB, and 7 to 10 percent seems good for EktaSpace. If you're working with one of these spaces, try turning the feature on and off to see if it's doing anything useful for you. Whatever you do, don't forget to turn it back off before you try to do any normal work; otherwise you'll find yourself producing excessively colorful imagery!

Blend RGB Colors Using Gamma

The Blend RGB Colors Using Gamma feature controls how RGB colors blend together. To see its effect, try painting a bright green stroke on a red background with the checkbox turned off, and then again with the checkbox turned on and the value set to a gamma of 1.0. With the checkbox turned off, the edges of the stroke have a brownish hue, as they would if you were painting with paint. With it turned on, the edges are yellowish, as they would be if you were painting with light. You can think of the behavior with the checkbox off as artistically correct, and with it turned on as colorimetrically correct. Permissible values are from 1 to 2.2.