Gamut Comparisons
Is sRGB or AdobeRGB (1998) Right for You?
© Richard Wagner 2004-2006 | WildNaturePhotos L.L.C.
The AdobeRGB (1998) working space deservedly enjoys wide popularity as a preferred color space for images destined for publication. This is because the gamut of the AdobeRGB (1998) color space is larger than the gamuts of applicable CMYK working spaces, such as US Sheetfed Coated. AdobeRGB (1998) therefore serves as a good intermediate when the ultimate CMYK output profile is not known. Of course, it is important to remember that when making conversions of images from AdobeRGB (1998) to a CMYK profile for offset printing, care must be taken in dealing with any out-of-gamut colors during the conversion. Here is a visual comparison of the AdobeRGB (1998) and US Sheetfed Coated profiles, as plotted in the Lab colorspace.
When initially working with scans from film or with RAW digital captures, the AdobeRGB (1998) working space may not be optimal. This is particularly true if your intended output device can print colors that are outside of the AdobeRGB (1998) gamut, as do the Epson Ultrachrome inkjet printers with Ultrachrome inks (2200, 4000, 7600, 9600, 9800, etc.). Limiting your working space to a gamut smaller than that of your intended output device will restrict the gamut of your output and will not take advantage of the full gamut that the printer is capable of printing. This limitation is greatest with sRGB, as shown below. When using an original that is in the sRGB color space, none of the colors that are within the wireframe Epson 9600 gamut plot that fall outside of the sRGB plot can be printed, even though the printer is physically capable of printing these colors.
The problem is not as bad with AdobeRGB, but a significant number of colors in the Epson 9600 gamut are still out-of-gamut in AdobeRGB.
Many of the yellows, greens and oranges of a custom Nikon LS-4000 scanner profile made from a Hutcheson-Velvia film target are outside of the AdobeRGB gamut. This gives rise to an undesirable situation where the image data captured by the scanner may be compressed into the smaller gamut of the working space, resulting in a permanent loss of color information. This scanner profile was generated by actually scanning the Velvia film target and building the profile using Gretag-Macbeth's Eye-One Match software. This is the most real world example as is possible.
Some blues also fall outside of the of AdobeRGB gamut.
This out-of-gamut problem can be avoided by using a wide-gamut working space. Note that it is necessary to begin in a wide-gamut color space. Converting from AdobeRGB to a wide-gamut color space offers no advantage. This can be accomplished by scanning to a raw scan and assigning a custom scanner profile, then converting to a wide-gamut space like ProPhoto RGB. With digital capture, just output the TIF image from the RAW processor using ProPhoto RGB or another wide-gamut color space.
Additional views of the AdobeRGB vs. HCT Velvia-Nikon scanner profile comparison are shown below.
As one might claim that the scanner profile is showing imaginary colors that don't exist on film, the actual measured reference values from the Velvia HCT target are plotted below.
The ProPhotoRGB gamut easily encompasses the entire Nikon-LS-4000 gamut as well as the Epson printer gamuts, making it an ideal working space for wide-gamut scans or digital captures. ProPhoto is also nearly identical to the profile that is used internally by the Adobe Camera Raw image processer (ROMM RGB).
Using a wide-gamut working space like ProPhoto RGB with scans and digital captures ensures that image data is not lost before output. It is the most versatile form to store images in when the final use is not known at the time of image preparation, and it is recommended by the ASMP and the Universal Photographic Digital Imaging Guidelines Working Group. The color accuracy of images in wide-gamut color spaces can be improved by using a 16-bit workflow. After the image is converted to the final destination color space (a printer profile or other output profile), the image can be safely reduced to 8-bit color.
Thanks to Steve Upton and ColorThink for the software that makes these comparisons possible!