Abstract
While luminance is the established standard for comparing the intensity of lights of different spectral compositions, it is known that it has several shortcomings in representing perceived heterochromatic brightness. With the widespread use of LED lamps with varying spectral properties, our goal was to evaluate how well luminance can capture the perceived intensity of light sources. We built a room with two seven-channel LED lamps placed on the two sides. The room was painted neutrally with RAL 7047 paint, and participants sat at its center. Observers were adapted to a D65 illuminant which led to a luminance of 58 cd/m2 at the wall. The illuminants were changed for 2 s during each trial, and 13 participants were required to choose the side on which illuminant appeared brighter. We used seven predictors for perceived intensity: luminance, radiance, intrinsically photosensitive retinal ganglion cell (ipRGC) excitation, the sum of CIE RGB channels (sumRGB), and their maximum of RGB channels (maxRGB). We also used weighted RGB combinations, with the weights estimated from earlier experiments on brightness judgments of objects. We created conflicting pairs of illuminants (e.g., left illuminant has higher luminance, but right illuminant has higher radiance), and compared which predictor performs best. When illuminants were conflicted between luminance and radiance, only 3 participants more frequently chose illuminants with higher luminance to be brighter than illuminants with higher radiance, and 54% of all judgments followed radiance rather than luminance. Compared to the weighted sum of RGB, luminance was only chosen in 37% of trials. Our results show that intensity judgments for lamps can be better predicted than by luminance. Even radiance seems to work better, and a weighted sum of CIE RGB that puts a greater emphasis on the short and long-wavelength part of the spectrum clearly outperforms luminance.