Toward a Perceptually Uniform Parameter Space for Filter Transparency

Abstract

Filter models of perceptual transparency relate to regularities in the retinal projections caused by light transmitting objects like clear liquids or glass and have been found to predict the color conditions for perceptual transparency more accurately than alternative models. An important but unsolved problem is how exactly the model parameters are related to the properties of the perceived transparent layer. We previously proposed a parametrization in terms of hue, saturation, overall transmittance and clarity of the filter that seems to capture important dimensions of the phenomenal impressions. However, these parameters are not independent and the corresponding scales are not perceptually uniform. Here, an invertible transformation of this parameter space is proposed that strongly mitigates these problems. This results in a more intuitively interpretable parameter set that seems well suited for the analysis of existing stimuli and the generation of transparent overlays with predefined perceptual properties. The latter property makes it suitable for graphics and visualization applications.