Abstract
We study the use of large reflectance sample sets to assess the color fidelity of light sources. We introduce three data sets covering a large fraction of the color space and a method to compute color fidelity over these sets. By applying this calculation to various source spectra, we show that some sources have significantly lower fidelity than would be predicted using the eight samples of the color rendering index. Structured and truncated source spectra are especially prone to this pitfall, whereas broad and smooth spectra are more resilient.
ACKNOWLEDGMENTS
The author is indebted to Michael Krames for helpful discussions and for reviewing the manuscript.