Purchase this article with an account.
Katja Doerschner, Huseyin Boyaci, Laurence T. Maloney; Estimating the glossiness transfer function induced by changing illumination and testing its transitivity. Journal of Vision 2004;4(8):322. doi: 10.1167/4.8.322.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The light reflected from a glossy surface depends on the reflectance properties of that surface as well as the spatial distribution of light sources in the scene. We asked observers to compare the glossiness of pairs of surfaces rendered under two different Debevec (1998) real-world illuminants (DIs). We used this data to estimate a transfer function that captures how perceived glossiness is remapped in changing from one DI to a second. We wished to determine the form of the transfer function, to test whether it was linear, and to test whether, for any set of three illuminants, the transfer function for DI1 to DI2 and the transfer function from DI2 to DI3 could be used to predict the transfer function from DI1 to DI3. We used Radiance to render spheres under 3 different DIs. We rendered each sphere as a stereo image pair for binocular viewing at 7 different specularity levels. The roughness of the surface was held constant, and the background was removed. We formed all possible pairings of spheres with 7 specularity levels under distinct DIs (3 × 7 × 7). On each trial, the observer saw the two spheres in a pairing side by side in a stereoscope. Each pairing was repeated 20 times. Three observers judged each of 2940 pairings in random order. They performed a 2ACF task, indicating which one of the two spheres appeared glossier. For each observer and each DI pair we computed a dominance matrix for perceived gloss. We obtained the contour of subjectively equal gloss which is the transfer function for that pair of DIs. These functions were close to linear. If the transfer functions were linear, we would expect the ratio of the slopes of the regression lines of perceived equal gloss of the first 2 pairs to predict the slope of the 3rd pair. Results from all 3 observers show the expected pattern. For 2 subjects the error of prediction was below 5% and for one subject it was 13%.
Grant EY08266 from the National Institute of Health; Grant RG0109/1999-B from the Human Frontiers Science Program
This PDF is available to Subscribers Only