The desire to get a better quality and a smaller file size is a usual dilemma in image archiving industry. One of traditional ways to reduce file size is adjusting a color encoding to characteristics of color reproduction device, while improving data precision is usually made by increasing bit capacity.By using a human-oriented color representation format that does not depend on color reproduction technique it has become possible to achieve better compression ratio at given precision compared to the most popular compression formats for high dynamic range images. The ability to store all visible colors and images with extremely high dynamic range is another benefit brought by the human-oriented approach. The accuracy of image data in this paper is measured in units based on experimental data on color visual threshold, a natural unit tied to human visual system capabilities. With a variable precision parameter it is possible to choose a desired data precision and to make a tradeoff between file size and data precision which is a useful option for digital image archiving.