Comparing lighting quality evaluations of real scenes with those from high dynamic range and conventional images

Thirty-nine participants viewed six interior scenes in an office/laboratory building and rated them for brightness, uniformity, pleasantness, and glare. The scenes were viewed in three presentation modes: participants saw the real space, and images of the spaces on a 17-inch computer monitor in both conventional and high dynamic range (HDR) mode. HDR mode allowed the high range of luminances in the real scene to be accurately reproduced, with maximum luminances more than ten times higher than those in the conventional images. For those participants who saw the images before the real spaces (the most relevant order for practical applications), the HDR images were rated as significantly more realistic than the conventional images. However, this effect was limited to scenes with relatively large areas of high luminance, which in this study was represented by scenes with windows and daylight. Ratings of the HDR images were significantly related to simple photometric descriptors of the images in the expected manner: brightness and glare ratings were positively correlated with overall and elevated luminance, and non-uniformity ratings were positively correlated with luminance variability. These results suggest that for evaluations of visual appearance of interior scenes featuring large areas of high luminance, the HDR method may be used as a surrogate for experiencing a real space both for lighting quality research, and in the design process.Trente-neuf (39) participants ont visionn\ue9 six (6) sc\ue8nes d?int\ue9rieur dans un \ue9difice contenant des bureaux et des laboratoires, puis les ont not\ue9es quant aux facteurs de luminosit\ue9, d?uniformit\ue9, d?agr\ue9ment et d?\ue9blouissement. Les sc\ue8nes en question ont \ue9t\ue9 visionn\ue9es selon trois (3) modes de pr\ue9sentation diff\ue9rents : les participants voyaient l?espace r\ue9el; les images de l?espace apparaissaient sur un \ue9cran d?ordinateur de 17 po en mode conventionnel; et ces m\ueames images \ue9taient pr\ue9sent\ue9es \ue0 cet \ue9cran en mode \ue0 grande port\ue9e dynamique (GPD). Le mode GPD permettait une reproduction pr\ue9cise de la gamme \ue9lev\ue9e de luminance de la sc\ue8ne r\ue9elle, comprenant des valeurs de luminance maximales de plus de 10 fois sup\ue9rieures \ue0 celles des images conventionnelles. Pour les participants qui voyaient les images avant les espaces r\ue9els (soit l?ordre de pr\ue9sentation le plus judicieux dans les applications pratiques), les images GPD ont \ue9t\ue9 not\ue9es par les participants comme \ue9tant consid\ue9rablement plus r\ue9alistes que les images conventionnelles. Cependant, cette incidence se limitait aux sc\ue8nes pr\ue9sentant des aires relativement \ue9tendues de haute luminance, soit, dans le cadre de cette \ue9tude, des sc\ue8nes incluant des fen\ueatres et de la lumi\ue8re naturelle. La notation des images GPD s?est r\ue9v\ue9l\ue9e \ueatre rattach\ue9e dans une large mesure \ue0 des descripteurs photom\ue9triques simples des images et de fa\ue7on pr\ue9visible : les notes pour la luminosit\ue9 et l?\ue9blouissement \ue9taient corr\ue9l\ue9es positivement \ue0 la hauteur (intensit\ue9) et \ue0 la globalit\ue9 de la luminance, et les notes pour la non-uniformit\ue9, \ue0 la variabilit\ue9 de la luminance. Ces r\ue9sultats laissent entendre qu?aux fins de l?\ue9valuation de l?apparence des sc\ue8nes int\ue9rieures comportant des aires de haute luminance \ue9tendues, le mode GPD peut \ueatre utilis\ue9 comme substitut de visionnement d?un espace r\ue9el, tant pour la recherche dans le domaine de la qualit\ue9 de l?\ue9clairage qu?au cours du processus de la conception.Peer reviewed: YesNRC publication: Ye