Color is one of the central elements of the photographic process. Rendering colors properly on a monitor or in a digital (or photographic) print is a complex subject that is extremely important to the success of the workflow. In order to work with color, you will need some grassroots knowledge about color itself and color models, and you will need to consider how to get your hardware to reproduce colors correctly. The monitor is the first and probably most important link in the chain and is the window on our images for most of the workflow.
This chapter addresses color, color models, and color reproduction while attempting to keep theory to a minimum.
Let’s start with the various color models (or modes) supported by Photoshop. A color model defines the way colors are described in a technical, mathematical way – i.e., the basic primary colors or primaries that make up a color and how those components are interpreted and arranged within color data. A color model also describes the amount of data required to record this information.
Photoshop supports several different color models, and the main color models used by photographers are:
Photoshop also provides additional color models that are rarely used by photographers, including Bitmap for pure black-and-white (bitonal) images, and Index mode, used primarily for Web graphics GIF files (if you can live with fewer than 256 different color values). Duotone is used with grayscale images and allows the addition of a second color, giving a print more depth and feel.
Color depth: In a color model, you can use either 8 bits (one byte) or 16 bits (2 bytes) to describe a single color channel, so you can store your image in 8-bit or 16-bit mode. Different bit depths are possible but are not supported by most applications. Using 16-bit color depth doubles the disk space necessary to store values, but it gives you more headroom when it comes to differentiating color values. It also allows for more precise calculations with fewer rounding errors. In 8-bit mode, the value of a single component can vary from 0 to 255 (using integers). (For the technicians among you: Integer values are used for 8-bit as well as for 16-bit data per color channel. Floating point numbers are used to describe some 32-bit formats.
In 16-bit mode, the single channel range increases from 0 to 65,535 (although Photoshop only uses 15 bits, reducing the maximum value to 32,767). We recommend using 16-bit whenever possible, but it is still common practice to use 8-bit values.
For most color management issues, it doesn’t matter which mode you use. You will need to reduce your image to an 8-bit format when you are producing output, as most printers are not capable of reproducing more than 256 different shades of a color. The various combinations of the three primary RGB colors add up to 16,777,216 (256 × 256 × 256) different reproducible colors. Our eyes can only differentiate between 120–200 hues of a particular color, depending on illumination, contrast, viewing distance, etc. 16-bit mode is nevertheless better for performing color optimization where color shifts, transformations, and resampling processes are performed.
All colors in the RGB color model are created from three primary colors: red, green, and blue. RGB is the most commonly used color model in digital photography, and we will perform our workflow mainly in this mode. RGB is an additive color model, meaning that the sum (addition) of all three basic colors at full strength (100 percent) adds up to pure white.
“0, 0, 0” defines black and “255, 255, 255” defines pure white. Pure white is rarely present in real-world photos.
The CIE L*a*b* color model separates colors (chroma, a + b) from the detail and brightness (luminance, L) in images. As with the RGB model, L*a*b* uses three basic components to describe a color: L (for Luminance), ranging from black (0, or no light) to white (100); and two color axes, a and b. The a-axis represents shades ranging from green to red (or magenta), and the b-axis represents blue to yellow shades.
The CMYK color model uses four primary colors to define a color: cyan (C), magenta (M), yellow (Y), and black (K). CMYK was designed for printing, where incoming light is reflected by the print.
CMYK is a subtractive color model, as each of the colored inks absorbs (subtracts) a certain component of the incident light. Figure 3-3 shows that mixing cyan and magenta gives you blue, and when you add magenta to yellow you get red. In theory, the combination of the colors C, M, and Y alone should be sufficient to produce black, but due to certain impurities in inks, they produce a dark muddy brown instead. To solve this problem, a fourth (black) color is added, and is called the key color (K for short).
Although CMYK is an important color model for industrial printing processes, it is not used much in digital photography. Though inkjet printers are technically CMYK printers (most are even CcMmYK with additional light cyan and light magenta inks), they provide the user with an RGB interface. Transformation from RGB to CMYK is performed by the printer driver as a background process.
We rarely use the CMYK color model in our workflow. Even when preparing images that requires CMYK for printing, you should stick to using RGB mode whenever possible and resort to an RGB-to-CMYK conversion at the very last step. After conversion, some additional sharpening and some slight increase in saturation may be required. Working on photos in CMYK mode has the following distinct disadvantages:
CMYK image files are larger than RGB files because they have four color values per pixel instead of three.
Some photo filters do not work in CMYK mode.
The CMYK color space usually contains fewer colors than most RGB color spaces. This means that when you convert an image from RGB to CMYK, you probably lose some colors. There is no way to retrieve lost colors if you want to use your image later for Lightjet direct printing on photographic paper or for a digital presentation on an RGB monitor.
Photoshop has a pure black-and-white (or bitonal) mode and a grayscale mode. In grayscale mode, the color of a pixel only describes a single (gray) value, so we usually work in RGB mode when we are producing monochrome images in order to preserve color information and to give our black-and-white prints some tint.
When using Photoshop’s bitmap mode, a picture has only two possible color values: black or white, no gray. Bitmap mode is rarely used in a photographic context, and most imaging tools and filters do not support it.
The HSB (Hue, Saturation and Brightness) or HSL (Hue, Saturation and Lightness) models are not explicitly supported by Photoshop, but are used in various tools such as the Photoshop Color Picker (Figure 3-35) or the Hue/Saturation tool. The hue is given using an angle from 0° to 360°. 0° (as well as 360°) corresponds to red, 90° to green, 180° to cyan, and 210° to blue. Saturation has a range from 0% (white) to 100%, while Lightness runs from 0% (black) to 100% (white).
In the Hue/Saturation and Saturation dialogs, saturation values are given in a range from -100 to +100. The same is also true for the Lightness slider. Zero then implies “no change”.
You can install a Photoshop extension from the Photoshop CD that provides HSB and HSL as color modes.
A color space is the total range of colors that real devices such as monitors or printers, or virtual devices such as a theoretical average monitor can record or reproduce. This range defines the gamut of the device.
Every real device has a unique color space, and even identical devices (same make and model) have slightly different color spaces, due to factors such as age and production tolerances. These differences increase with variations in user-selected hardware or software settings, such as different monitor resolutions, different printer inks or papers, or even different brightness settings on a monitor.
To improve color consistency and applicability, the International Color Consortium (ICC) and some other companies (Adobe, Kodak, Apple, etc.) have defined virtual color spaces representing the gamut of virtual rather than real devices. We will discuss the advantages of virtual, standardized color spaces later (e.g., Adobe RGB, Apple RGB, sRGB, ...). One advantage of these artificial RGB color spaces is the fact that they are gray neutral, implying that any equal amount of R, G, and B will result in a neutral gray tone. This is not the case for most real device color spaces.
For more information on the ICC and color management in general, see www.color.org.