 |
November 20, 2009 |
  |
Created and Maintained by: The Photoimaging Information Council |
|
|
by Andrew Rodney |
|
When viewing digital images in Adobe Photoshop, it’s often difficult to see the forest from the trees. Most users don’t realize that all they are viewing is a huge collection of solid colored pixels, each represented by a set of numbers as illustrated in figure 1. Colored pixels defined by numeric values are the only way a computer can produce what appears to be a continuous tone photograph. Your digital darkroom is nothing more than an adding machine. All Photoshop can do is alter the numeric values of these pixels.  Figure 1. The enlargement seen at 1600% zoom shows a single pixel with a numeric value of R103/G 52/ B 39. Every solid pixel in this image has an associated color number. Only when zoomed out does this mosaic of solid colors produce what appears to be an image.
When viewing an image on-screen that appears too dark, you’d never turn up the brightness on your display to correct this problem. If your prints appeared too cyan, you could in theory add a red filter to the light source under which you viewed the print, but this wouldn’t solve the colorcast issue. Instead you would edit your images by changing the pixel values until you produce the desired color appearance. You expect the color and tone you see on your display and the color and tone of your print to match. Otherwise you have to make multiple edits to your images and multiple prints until you produce the color you desire: a time consuming and expensive process. How do you know the numbers associated with the pixels define that image correctly? The only way to remove such ambiguity is to utilize color management, which is really nothing more than the management of numbers! Color management is necessary when working with digital images because numbers alone do not fully describe the appearance of color, but rather a partial ingredient for mixing colors. This is analogous to receiving a recipe for a meal that lists all the ingredients but not the amount of ingredients in cups, grams, teaspoons or other units of scale. Many users assume they can work “by the numbers” alone. Imagine a single solid color displayed in Photoshop. Let’s use Red 240/Green 78/Blue 98 as an example. What color is this and how should it appear? You can presume this is a shade of red as the Red numeric value is so much higher than green and blue. But how red is this red and will this shade and tone of red reproduce as you expect? The missing information is the scale of the numbers much like the scale of the ingredients in our recipe. In order to correctly associate the numbers with a scale, it is necessary to define what is called a Color Space. The term color in this context is obvious. The term space simply tells us where the numbers fall within a defined scale. This scale is based on the visible colors we humans can see. All the colors we are able to see, the visible light spectrum, defines the color space of human vision. Within that scale, all other color spaces can be defined. If you have ever heard of sRGB, Adobe RGB (1998) or other named color spaces, you are simply being told the scale of specific numbers in a defined color space. For example, the most saturated green you can produce numerically is always Red 0/Green 255/ Blue 0. This is true for any color space however, where that specific color lies within the color space of human vision is different as seen in figure 2. The horseshoe shaped plot seen here is the color space of human vision and within that color space is where the most saturated green of both sRGB and Adobe RGB would fall. The volume of a color spaces is known as the Color Gamut of that color space. If I simply specify Red 0/Green 255/ Blue 0 without an associated color space, you would have no idea how green this green really is. By specifying numbers and a color space, you can now reproduce that color exactly as I can. One way to describe a color space is by using ICC profiles. These profiles define the scale of the color space, as seen in figure 2. ICC profiles give numbers a definition or scale within human vision.  Figure 2. The horseshoe shaped plotting of color represents the visible colors humans can see. This is the gamut of human vision. With this as our total scale, it is possible to plot any other color space and seen here is sRGB and Adobe RGB (1998). Note that the most saturated green can numerically be defined as G 255 but the scale is different in each color space.
There are two requirements necessary to ensure the numbers you manipulate produce the desired color appearance both on-screen and when printed. First, you need an ICC device profile for your display to describe how it reproduces color and define its unique color space. You also need an ICC profile to define the color space of your documents; the scale of these numbers as they relate to human vision. R240/G78/B98 in two different color spaces, for example sRGB and ProPhoto RGB do not produce the same color even though both share identical numbers. In Figure 3 I have placed sample points on several colors in two documents, in two different color spaces. In this example, the color appearance is the same yet the color numbers are different. This is to be expected because the colors are in different color spaces. Photoshop needs each document to have an embedded ICC profile so it can display these numbers, as they should appear. Color management instructs Photoshop how to properly display and output the numbers and thus colors you expect. ICC profiles work in many applications where color appearance and color matching is critical.  Figure 3. The same appearing colors produce different numeric values because in this example, you are viewing two different color spaces. Since Photoshop knows the scale of the numbers based on the embedded ICC profile in the document, it can properly preview these numbers using the display profile, which defines the monitors color space.
If you wish to output an image to a certain printer, you need an ICC profile that defines how that output device behaves, defining its unique color space. Photoshop then does its mathematical magic by altering every pixel value to produce a new set of numbers optimized for that printer. Just as the same numbers in sRGB and ProPhoto RGB produce a different color, the same set of numbers sent to 2 different printers would produce two different colors. When Photoshop knows the color space of your output device, it can produce the correct color values, and it can simulate on-screen how these colors should appear. This on-screen simulation is known as a Soft Proof and it is an incredibly powerful feature. You are able to see how your image will appear on screen before you spend the time and money to actually make a print. You need ICC profiles for all the devices you plan to utilize; displays, scanners, printers and so forth. You need ICC profiles to define the color space of images you will edit. However ICC device profiles know nothing about images, only the device behavior they define. Therefore, human intervention is usually necessary. Just as ICC profiles (and Photoshop) only understand color one pixel at a time, it is up to you to view the pixels in context and decide how you want your images to appear. In order to properly handle these images, you need to calibrate and create an ICC profile of your display on a regular basis. Displays behave inconsistently, even from identical models and their behavior changes over time. You should profile such devices on a monthly basis ideally using an instrument (usually with a device known as a colorimeter). The same set of numeric values should produce the same color appearance every time you view them. You should receive files from your scanner or digital camera with embedded ICC profiles that correctly define the image’s color space. If you are sending files to a lab for output, ask for an ICC profile so you can properly view your images as they will appear when output. Never alter your calibrated display to match a reference print or file from a lab; this will ruin Photoshop’s ability to produce an accurate soft proof. When properly configured, Photoshop will correctly display and output your numbers…I mean images.
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© 2002 - 2009 Take Great Pictures |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
