Museum Lighting: Adjusting The Illuminant #1916 Museum Lighting: Adjusting The Illuminant   I. Abramov1, J. Gordon2, M. Scuello 1, and S. Weintraub 3 Applied Vision Institute/Brooklyn College CUNY1 Hunter College CUNY2 Art Preservation Services3, New York. Brooklyn College ABSTRACT MUSEUM LIGHTING: ADJUSTING THE ILLUMINANT. ((I. Abramov1, J. Gordon2, M. Scuello 1, and S. Weintraub 3)) Applied Vision Institute/Brooklyn College CUNY1, Hunter College CUNY2, Art Preservation Services3, New York. Purpose. We had shown earlier that viewers prefer to look at art works lit by higher color temperatures than the typical lighting of 3000 K at 200 lux. The preferred illuminant was about 3600 K at 200 lux, but the illuminants used differed by steps of about 400 K. To refine those findings, observers now continuously adjusted color temperature and repeated the settings at each of four illuminances. We also tested the hypothesis that the preferred illuminant would appear neither warm nor cool. Methods. The observer looked at a “painting” hung on a matte-gray wall lit by overhead banks of lamps whose combined value could be adjusted continuously between 3000 and 4400 K while keeping illuminance constant. This was repeated for paintings varying in subject matter and color. On different sessions the illuminance value was one of 50, 200, 500, or 2000 lux, achieved by wearing goggles with Inconel neutral density filters. Each trial began with the illumination set, randomly, to one of 3000, 3600, or 4400 K; observers adjusted color temperature until they were satisfied they found the “best” illumination for each painting. In a separate study, using the same procedures, observers viewed a white reflectance standard and adjusted the illumination until it appeared neither warm nor cool. Results. The chosen illuminant for viewing paintings was about 3800 K, independent of intensity from 50 to 2000 lux; settings were largely the same across paintings and across eight observers. A different group (N=8) chose approximately 3700 K as neither warm nor cool, independent of intensity. These illuminants are not “achromatic” but correspond to a dominant wavelength of about 580 nm at 35% purity. Conclusions. Illumination in many museums is suboptimal for aesthetic enjoyment. Preferred illuminant is one that is neither warm nor cool, rather than achromatic. We calculate that for the preferred lighting, the small addition to short-wavelength energy should have minimal impact on preservation. CR: P. Support: National Park Service grant MT-2210-8-NC-21, and PSC/CUNY grant, 62306-00-31. INTRODUCTION There are two constraints on the way in which artifacts displayed in museums are lit: conservation and aesthetics. Typically the lighting has a color temperature of about 3000 K and illuminances between 100 and 300 lux. Current practice is often based on a canonical set of illumination functions that are said to yield the most pleasing illumination in the workplace, assuming that something neither warm nor cool is best (Kruithof, 1941). At 50 lux the range is 2200 to 2500 K while at 2000 lux it is from 3500 to more than 10000 K; the midpoint of the range at 200 lux is the widely used 3000 K. Earlier we had simulated a miniature museum environment: observers either compared the appearance of the same painting under different illuminants in adjacent rooms or rated the appearance of paintings seen one at a time under different illuminants. Illuminance was approximately 200 lux; color temperatures ranged from 2500 to 7000 K, representing “white” lights from early morning sunlight to light from a blue, cloudless northern sky. Observers preferred a color temperature of about 3600 K. The result held true across a large variety of paintings. It was also unaffected by pre-adapting the visual system to different illuminants (2800 K, 3600 K, 5800 K). See Fig. 1a,b,c,d. Our previous work used one illuminance and varied color temperature in coarse steps. Here we extend our work by using lights whose color temperature could be continuously adjusted and also varied illuminance over a large range that includes the typical level in museums. We also tested the hypothesis that viewers of artworks prefer lights that appear neither warm nor cool. METHODS The observer looked at a “painting” hung on a matte-gray wall lit by two overhead banks of lamps whose combined value could be adjusted (key-press on remote keypad) continuously between 3000 and 4400 K while keeping illuminance constant. One bank contained 50W, 12V, MR-16 lamps (Solux) nominally rated at 4700 K; the other bank’s lamps, 35W, was nominally 3000 K. In each session 14 paintings varying in subject matter and color were presented one at a time. On different sessions the illuminance value was one of 50, 200, 500, or 2000 lux, achieved by wearing goggles with Inconel neutral density filters; these sessions were repeated. Sequences of intensities and pictures were randomized. Each trial began with the illumination set, randomly, to one of 3000, 3600, or 4400 K. Observers (N=8) adjusted color temperature until they were satisfied they had found the “best” illumination for each painting. In a separate study, using the same procedures but with a white reflectance standard (Photo Research) hung in place of a painting, a different group of observers (N=8) viewed and adjusted the illumination until the neutral white reflector appeared neither warm nor cool. A session used only one intensity (random order for each observer); in a session a total of 12 settings was made, of which the first three were practice trials. CONCLUSIONS ü      Most museums use sub-optimal illuminants. ü      Observers consistently prefer higher color temperatures between 3500 and 3900 K. ü      Preference is independent of intensity over any realistic range of illuminances. ü      Preference is also independent of state of adaptation of the visual system across a range of “white” lights from tungsten illumination to average daylight. ü      Observers prefer to look at paintings under a light that appears neither warm nor cool rather than a light that appears achromatic. ü      A light that appears neither warm nor cool is one that does not evoke any sensation of redness. RESULTS & DISCUSSION 1) Preferred Illuminant for Paintings The chosen illuminant for viewing paintings was about 3800 K, independent of intensity from 50 to 2000 lux; settings were largely the same across paintings and across eight observers (Fig. 2a). Consistency of settings was examined by comparing results from the two sessions at each illuminance; there was no significant change (Fig. 2b). There was no correlation between illuminant settings and colors or subject matter of the paintings. Adjustment procedures are prone to “anchor” effects; we examined this by averaging separately trials that began at each of the three starting color temperatures; there is a small effect of starting point on final setting (Fig. 2c). We calculate that shifting the illuminant to 3800 K, while keeping illuminance constant, has minimal effect on irradiance; there should be minimal consequence for preservation. 2) Neither Warm nor Cool Illumination The group’s mean choice of an illuminant that was neither warm nor cool was approximately 3700 K, independent of intensity from 50 to 2000 lux (Fig. 3a). As with the settings chosen as best for viewing paintings, there was a small relationship between final setting and starting point for each trial (Fig. 3b). The mean setting of 3700 K corresponds to a light with dominant wavelength of about 580nm at a purity of 35%. Seen as a stimulus patch, such a light appears quite yellow. None of the observers adjusted the lights towards the higher color temperatures that appear achromatic. In a separate study four observers scaled the apparent warmth or coolness of flashes of equi-luminant monochromatic lights; the warm-cool transition was between 560 and 580 nm (Fig. 4a). The same observers also scaled the sensations of hue and saturation elicited by the same lights (Fig. 4b). Warmness was completely predicted by the perceived redness of each light (Fig. 4c). One comparison is with percent redness seen at each wavelength and one is for redness re-scaled to include the drop in saturation as wavelength decreased from 650 to 550 nm. Both curves agree with perceived warmth, suggesting that a light eliciting any sensation of red will appear warm even if the red is desaturated.