INTRODUCTION

Many users of high quality colour printers complain that the colour reproduced on hardcopy does not always match the original softcopy images previewed on a self-luminous display. Such complaints arise even when colour management systems are used in the reproduction technology. There are several possible causes for these problems such as the instability of the colour reproduction in the input and output devices or the inaccuracy of the colour characterization. This report focuses on another potential problem - the difficulty of determining the state of chromatic adaptation of the observer.

The use of colour management systems to reproduce colour images is commonplace in colour imaging industries and is also found in the consumer marketplace. For example, in the pre-press industry hardcopy images are often reproduced and viewed on a computer monitor as a soft proof prior to making a final print. Because colours appear differently when viewed under different viewing conditions, and viewing configurations of various media differ from one another, the colour appearance of the reproduction of the image on the monitor does not generally match that of the original hardcopy print. In this case, to achieve an accurate cross-media colour reproduction, one needs a colour appearance model (CAM), which is capable of predicting the corresponding colours from one set of viewing conditions to another. In general, CAMs comprise three parts: a chromatic adaptation transform, perceptual correlates, and a uniform colour space. Of these three, the chromatic adaptation transform plays a vital role for accommodating between two different illuminations in two media viewing conditions.

The chromatic adaptation transforms (CATs) used in most CAMs assume that observers are fully adapted to a given set of viewing conditions. That is, the chromatic adaptation transform is designed to reproduce the appearance as seen by an observer who is fully adapted to an original viewing condition after that observer has fully adapted to a second viewing condition. The requirement of complete chromatic adaptation allows for more rigorous and reproducible results. Such a requirement is perfectly appropriate for laboratory research and for industry use by informed and trained professionals. This requires, for example, that observers view a softcopy image in an environment where the display luminance is higher than any ambient illuminant. Further, observers must take several minutes to adapt to the softcopy viewing conditions.

Unfortunately, the condition of complete chromatic adaptation usually does not occur in the consumer market and in more casual industrial use. Instead, these users tend to view softcopy in a room with sufficient ambient illumination to allow comfortable viewing and examination of hardcopy imaging. Further, users often desire to compare hardcopy and softcopy images using rapid successive binocular observations. To put it simply, they glance back and forth between the softcopy and the hardcopy display.

In 1998, Technical Committee 8-04 was formed in CIE Division 8 (Image Technology), "to investigate the state of adaptation of the visual system when comparing softcopy images on self-luminous displays and hard copy images viewed under various ambient lighting conditions". The focus of this report is to determine how colour imaging experts can best accommodate the desires and practices of these more casual observers. TC members realized that the results would necessarily be more approximate than those typical of CIE reports. By design, our experimental conditions did not achieve the degree of rigour typically associated with colour science. Despite this, we were able to achieve several useful results. We determined that accounting for mixed and incomplete chromatic adaptation (the terms will be defined below) produced more accurate results in colour appearance than not accounting for them. We determined a mathematical model for chromatic adaptation that accounts for our experimental results and integrates well with other CIE work on appearance models. Finally, we determined appropriate parameters for the chromatic adaptation model.