Spatial brightness
An experiment was conducted to compare spatial brightness at photopic levels under two types of incandescent lamp, one with a clear-glass bulb and the other with a blue-glass bulb. This was done following a paper by Joe Lynes suggesting the brightness of daylight is undervalued by conventional photometry, the blue-glass bulb being marketed as a ‘daylight simulating’ lamp.
A side-by-side matching task was used: for equal brightness, the blue-glass lamp required significantly lower illuminance than the clear-glass version, which was in accordance with Lynes’ proposal.
The experiment was continued but using a broader range of lamps. This focused on the breadth of lamp spectrum, ranging from low pressure sodium lamps, high pressure sodium lamps to full spectrum fluorescent lamps, with warm white fluorescent used as the reference source.
Compared to the warm white lamp, the HPS and LPS required significantly higher illuminances for equal brightness, and the full spectrum lamp significantly less.
These results show that changes in lamp spectrum affect brightness, but do not, alone, provide a metric to extend the application.
Results from Fotios and Levermore 1997 were used to test and develop three potential models for spatial brightness.
The first model considered colur gamuts, the area of the polygon created in colour space by the chromaticity of a series of test colours when lit by the specific lamp. Following proposals by Schanda and Thornton, this study considered extending colour gamuts from 2D to 3D. The best fit for these data was Cone Surface Area. This cone was defined as having the base as determined for gamut area in u’-v’ chromaticity space and height as determined by the w’ chromaticity.
This model was subsequently used by others as a potential metric for colour rendering: despite that not being the original intention, it worked reasonably well in that role.
In the 1990’s there was strong promotion by some that spatial brightness at photopic levels could be explained by the S/P ratio (in other words, a rod contribution to photopic brightness) with lamps of higher S/P ratio producing greater brightness.
The literature, however, did not clearly support a rod contribution to brightness - but did support a contribution from the short-wavelength sensitive (SWS) cones.
Hence an alternative model was created using the SWS/P ratio rather than the S/P ratio and this provided a better fit to test results.
The third model started by considering simple opponent colour models of brightness (e.g. Thornton’s Brightness Meter and Guth & Lodge), and then developed a variant based on the 1997 test data.
A common approach to developing a model is to design an experiment to test a particular theory. There is always a risk, however, that the theory is supported by coincidence of an experimental bias. A robust validation is to test that model on independent data. A literature review was carried out to gather such data.
The data did not appear to fit a model based on either CCT, CRI, gamut area nor chromaticity alone: what did appear to provide a consistent prediction was a two-parameter model, the simultaneous consideration of CCT and CRI, but this paper did not attempt to develop a mathematical model for this relationship.
An experiment was designed to test the S/P ratio as a metric for spatial brightness at photopic levels, or more specifically, the brightness lumens model: P(S/P)0.5. Two luminances and two S/P ratios were used in all four possible combinations, the precise values chosen to include a low luminance and high S/P ratio combination that was predicted to be equally bright as a high luminance low S/P ratio combination. Two different procedures were used.
We concluded that “These data suggest that spatial brightness perceptions at photopic light levels are unrelated to the S/P ratio of the illumination.” Perhaps this might have be better interpreted as S/P ratio alone is insufficient to characterise the effect of spectrum on spatial brightness.
This is a continuation of the 2001 literature review, with the number of case studies increased from 21 to 70.
A focus of this review was to identify those studies which provided credible evidence, as demonstrated by inclusion of experimental precautions (counterbalanced or randomised orders), checking (null condition trials), verification (the use of multiple procedures) and clearly reported findings. Of the 70 studies, only 19 were considered to provide credible evidence.
This experiment was carried out to directly (or, as near as possible) repeat the Berman et al 1990 study of spatial brightness which had demonstrated that lighting of higher S/P ratio appeared brighter. Thus, two lighting conditions had identical chromaticity but different S/P ratio. A third condition was added, to allow comparison of different gamut areas but constant S/P ratio.
Three test procedures were used: a repeat of the discrimination procedure used by Berman et al, a matching procedure and a second discrimination procedure. The experiment also included the null condition absent from Berman et al.
It was concluded that while lighting of higher S/P ratio was brighter, the S/P ratio alone was insufficient to predict spatial brightness. A metric for the chromatic contribution is also needed, this being provided by gamut area in the current work.
Many studies attempt to validate the Kruithof curve, an alleged relationship between CCT and illuminance that defines the pleasant combinations. However, this critical review of Kruithof-like studies revealed that the relationship is not supported by credible data. Specifically, it did not suggest any consistent effect of CCT on brightness or pleasantness.
Fotios S. A revised Kruithof graph based on empirical data. Leukos, 2017; 13(1); 3-17.
