Accuracy and Reversibility of Color Conversions

The quality of the color reproduction of an image converted from one colorspace to another depends on the following factors:

» The overlap between the gamut of the source device and the gamut of the destination device

» The quality of the color profile used (such as the number of samples used to generate the conversion transform)

» The accuracy of the color profile

» The rendering intent selected

Devices with similar gamuts will produce the smallest color distortion. Typically, converting from the colorspace of one monitor to that of another monitor results in no noticeable loss of color quality. However, converting from an RGB colorspace to a CMYK colorspace will typically result in changes in the color from the original. This loss is caused by the fact that monitors typically have larger gamuts than printers. For photographic images, this loss is typically small because most naturally occurring colors can be produced equally well on monitors and printers. For computer-generated images, the loss is typically larger since there are many colors representable on a monitor that cannot be achieved with ink on paper.

Printer profiles are usually constructed with a large number of sample points for converting colors to the printers colorspace. However, to make the profiles reasonably small, far fewer sample points are usually provided for converting from the printers colorspace. Therefore, color conversions using printer profiles as the source colorspace should be avoided whenever possible.

Because of the inherent inaccuracy in the color conversion process, converting from one colorspace to another and back should also be avoided where possible. Images for processing should be converted to a common colorspace, processed, and converted to the destination. Choice of the common colorspace depends on the colorspace of the images involved, the type of images involved, the quality of the color profiles, the destination colorspace, and the operations to be performed.