Color Study Challenges CRI

Below is one of my contributions to the January issue of tED Magazine, the official publication of the NAED. Reprinted with permission.

The room and objects used for the color study.

The room and objects used for the color study.

A study recently published in the journal Lighting Research & Technology explored human preference related to light source color. The results challenge the ability of the color rendering index (CRI) metric to predict human preference and the ability of today’s energy-efficient lighting to satisfy preference.

Color vision is highly dependent on three factors: the eye, the object viewed, and the spectrum of the light used for vision. CRI is the traditional metric used to evaluate how well light sources render colors. It expresses how accurately a source renders colors compared to an ideal reference source. For many commercial applications, the industry recommends a minimum 80 CRI.

CRI has always had its limitations, which the proliferation of LED lighting accentuated. In 2015, the Illuminating Engineering Society proposed TM-30, a new method for light source color evaluation.

TM-30 proposes three tools:

• Fidelity Index (Rf) (0-100 scale): an alternative to CRI that uses 99 color samples instead of CRI’s 8-14.
• Gamut Index (Rg): a measure of average color saturation. This metric complements the fidelity index.
• Color vector and distortion graphics. These valuable graphics depict hue and saturation changes, or gamut shape. They show specifically what colors have increases or decreases in saturation. Colors that have an increase in saturation will visually pop.

Researchers at the Pacific Northwest National Laboratory (PNNL) used TM-30 and CRI to investigate subjective impressions of color quality, specifically normalness, saturation and preference. They hypothesized:

• As Rf increases, colors would be judged as more normal.
• As Rg increases, colors would be judged as more saturated.
• Greater saturation, specifically reds, are more preferred.

Twenty-eight people with varying ages and gender evaluated a variety of objects under different lighting conditions. The object colors presented a range of hue, saturation and lightness. The room was lighted to 20 footcandles and a constant 3500K correlated color temperature. The light sources had varying combinations of fidelity and saturation.

As expected, Rf is a good predictor of how normal colors appear. Rg is a good predictor for saturation. However, alone they are not good predictors of preference. A light source with the same Rf and Rg values can produce different impressions. That’s because Rg is an average indicator of saturation, and participants specifically showed a distinct preference for saturated reds.

The preference for saturated reds is interesting in light of the fact most energy-efficient lighting does not saturate reds. Additionally, many of the most favored sources in the study had CRI values under 80. Because CRI is biased against saturated reds, manufacturers engineer light sources that achieve a good CRI but less-preferred gamut shapes.

Click here to learn more.

Leave a Reply