This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility: The candela, the international unit of light in use for almost a century, forms the basis of photometry. According to a new study, measurements derived from it systematically misjudge the brightness of colored light sources and depart markedly from human perception.
An international team led by perception researcher Karl Gegenfurtner at Justus Liebig University Giessen (JLU) has found a surprisingly simple rule for how bright colored surfaces actually appear: Perceived brightness can be predicted almost completely by taking only the highest of the three weighted color components: red, green and blue. The rule accounts for more than 95% of the judgments observed and outperforms all established photometric models. "The candela is the only SI unit that rests not on physics alone but on human perception," Gegenfurtner says.
That the quantities derived from it diverge so clearly from perception in the case of colors has consequences for the labeling of lamps, the calibration of displays and the planning of lighting in offices, on streets and in homes. For the observers in the study, published in Proceedings of the National Academy of Sciences, saturated colors appeared considerably brighter, and blue light contributed more strongly to perceived brightness than photometric measurement predicts. Gegenfurtner traces the discrepancies to the assumption underlying the standard, which was established in the 1920s: that the brightness of a color mixture equals the sum of its parts.
Measurement by classical brightness matching is also slow, difficult and unreliable among observers. Working within the European Research Council project Color 3.0, the Giessen team took a different route and simply asked observers to sort colored patches from bright to dark. The method proved remarkably robust.
Observers ranked 144 colors, and their orderings were nearly identical when repeated more than six months later. They remained just as stable when the same people repeated the task at home on their own uncalibrated monitors and when a further 486 participants completed it online. For the first time, the brightness of different colors could be measured on a large scale.
With these data, the team tested a wide range of candidate models, including luminance, radiance and established color appearance models such as CAM16. "None of these models captured the pattern. What worked in the end was strikingly simple," Gegenfurtner explains.
Extract — continue reading at the source.
