Color Encodings: sRGB and Beyond

Most digital imaging products aimed at the open desktop market are designed to produce and accept image data in the sRGB color encoding. sRGB is defined with respect to the response of a reference CRT display. As a result, the image data can be easily interpreted, and can be directly displayed on a typical CRT without the need for additional color transformations. Thus, sRGB simplifies the workflow for softcopy-based viewing, editing, and image sharing. One intent of sRGB is to standardize the way in which images are stored and communicated in consumer digital imaging systems, thereby improving the interoperability of these systems. However, for non-CRT-centric applications, limi­ tations associated with current sRGB-based workflows can negatively impact process complexity and image quality. This paper will discuss the pros and cons of several differ­ ent approaches that have been proposed to overcome these limitations. Color Gamut Considerations sRGB Color Gamut Issues One issue that is important in many digital imaging systems is the ability to fully and optimally utilize the color gamut of the output media. Because the sRGB color encoding is specified relative to the response of a standard CRT, the colors that can be represented are limited to those within the color gamut of this standard display. As a result, storing images in sRGB can limit the capability of an imag­ ing system to accurately reproduce colors outside the sRGB gamut. Film, scanners and digital cameras can both capture colors well beyond the sRGB color gamut. Therefore, it is necessary in sRGB-centric workflows to color render the captured image data into the sRGB gamut. Likewise, many output devices have color gamut shapes different from that of sRGB, with the device color gamut extending beyond the sRGB color gamut in some regions of color space, and the sRGB gamut extending beyond the device gamut in other regions. For example, as illustrated by the gamut slices in Fig. 1, photographic printers employed in digital photo­ finishing typically have a larger gamut for dark colors. And a smaller gamut for light colors. As a result, when printing sRGB images it is necessary to gamut map from the sRGB gamut to the actual output device gamut. If the gamut mapping out of sRGB is designed to be complementary to the color rendering into sRGB, this practice can produce acceptable results. However, in many applications, the input and output processing will be inde­ pendent, and therefore inconsistencies between proprietary color rendering and gamut mapping (or even gamut clip­ ping) algorithms can often produce less than optimal results. Even in closed systems, where both the input and output processing can be coordinated, this co-optimization of the color rendering and gamut mapping can introduce a significant amount of additional computation and compli­ cation. These gamut restrictions are even more significant when color enhancements are applied to boost the colorfulness of an image, or where it is desired to accurately specify special colors (e.g., PANTONE∗ colors) that are outside the sRGB color gamut. 100 50 0 50 100 0 20 40 60 80 100