Six Things You Should Not Do With CIELAB

CIELAB has served us well. It has provided a reliable and unambiguous way for a brand owner to uniquely specify a target color. It has enabled spectrophotometers to assist through the workflow by providing standardization of color. ICC profiles and G7 are based on CIELAB. By using color difference formulas based on CIELAB, we have an objective way to determine if a printed product is in spec.

But, CIELAB is not without foibles. Its development occurred without a full knowledge of how it would be used. Some of the decisions that went into CIELAB have had far-reaching and underappreciated consequences. As a result, there are some expectations that we have for CIELAB that are questionable.

This paper describes a number of the problems with CIELAB by looking at where it does not perform well. There is a subtle undercurrent in the paper that the industry needs to start considering how to replace CIELAB with a more modern color space that does share CIELAB’s issues.

Things you should not do

1. Never call the a* axis red/green.
2. Don’t assume that colors opposite one another are complementary colors.
3. Don’t assume that two colors with the same CIELAB hue angle have the same perceptual hue.
4. Never, ever compute the color difference between two CIELAB values computed with different illuminants. Just don’t do it.
5. Don’t assume that CIELAB is perceptually linear.
6. Don’t trust any color difference formula (even ΔE00) when the illuminant is significantly different from D65.