Abstract
We have investigated the effect of colour temperature and illuminance on the visual response. Since Kruithof's (1941) work on general illumination, it has been proposed that users would find ‘pleasing’ high correlated colour temperature (CCT) illumination at high illuminance and low CCT illumination at low illuminance, although it is not unanimously accepted. Here, we question whether the pleasing sensation comes from facilitation of visual performance or from subjective appraisal. We have conducted experiments at various illuminance levels and various CCTs including performance tasks and subjective scaling of illumination quality. The change of illumination was obtained using adjustable LED clusters with high colour rendering index (R a > 90). Whereas the performance tasks yield results which depend on luminance and only slightly on CCT, observers reported that high CCT illuminations look brighter than low CCT illuminations. We investigate the effect of the various spectral power distributions on the intrinsically photoreceptive retinal ganglion cells and on the pupil response.
Acknowledgements
We would like to thank Corinne Talotte from the Direction de la recherche de la Société Nationale des Chemins de fer Français for initiating this study, Alain Bricoune and Frantz Dennery from LedToLite for providing the LED illumination system, Eric Castet and François Vital-Durand for counselling us on how to choose reading texts, Danièle Dubois and Gaëlle Delepeau from Laboratoire d'Acoustique Musicale for guidance in preparing the questionnaire, Jean Le Rohellec for photographs and comments and Kenneth Knoblauch for comments on the manuscript.
Notes
Notes
1. The English version is a translation from French of the subjective scales proposed to evaluate light quality: Non éblouissante–Eblouissante; Sombre–Lumineuse; Froide–Chaude; Rendu des couleurs artificiel–Rendu des couleurs naturel; Crépusculaire–Claire; Triste–Gaie; Fatigante–Relaxante; Inconfortable–Confortable; Désagréable–Agréable.
2. Introducing a reference CCT T
cp,o would yield the model
with α = 0.821918, β = 0.000467, m = 0.31, n = 0.96 and T
cp,o = 4. It only slightly modifies the prediction.