| dc.description.abstract | Light has been researched by many fields such as physics, philosophy, psychology, pictorial arts and psychophysics. The latter brought about the concept of the physical light field which can be summarized as a function that measures the distribution of light in a certain space on every possible point (x,y,z,). An aspect of the physical light field is illuminance flow which is a robust indicator of the light field computed through a series of algorithms. On the other hand, research has shown that people also quite clearly observe the visual light field. This concept is different from the physical light field because the visual light field is defined in, for example, the empty space between objects and therefore has no physical existence. With this information, a research was conducted investigating the visual light field within paintings. The perceptual results obtained from that research were used in the present paper as material to be compared with illuminance flow calculations of the same paintings. Previous research suggested that there might be a relation between the scale on which the algorithm calculates the illuminance flow and the perceptual data. This would provide us with a ‘perfect scale’ on which the algorithm operates. In the present paper, the relation between scale selection and perceptual data was investigated after which the question arose whether there is also a relation between the performance of the participants in the perceptual experiment and the performance of the algorithm. The algorithmic calculations were executed using Mathematica, and the scale of the algorithm was varied over all the calculations. The results of the algorithm were documented in SPSS and correlation analysis was conducted. The results showed no evidence for a perfect scale because there was no relation between the calculated orientation of the algorithm and the average orientation of the perceptual data. There was however a significant relation between the standard deviation of the perceptual data and the confidence of the algorithm which suggests there is a relation in the performance of both. This led to the conclusion that the harder it is for participants to estimate the illuminance flow in a painting, the lower the confidence of the algorithm becomes. Because the first hypothesis was rejected, suggestions for future research have been made mainly regarding the constancy of the stimuli. | |
