Abstract
Viscosity is a material property that is unique to fluid. In physics, viscosity is described as a parameter for the motion of fluid. This means that motion information should contain information of viscosity. In psychophysics, human observers can veridically estimate fluid viscosity just by looking (Kawabe et al., 2012, VSS; Fleming et al., 2012, VSS). Can the visual system estimate fluid viscosity solely from motion information? To examine this question, we created a novel motion stimulus, called the simulated motion field (SMF), which consisted of an array of patches. The carrier pattern of each patch was a low-pass filtered white noise, and moved within a stationary circular envelope. The array of carrier movements simulated optical flow patterns extracted from CG animations of scenes including moving liquids. We obtained the following results. (a) SMF gave observers a vivid impression of dynamic liquid. (b) In accordance with an increase in the kinetic viscosity of the original CG animation, the viscosity rating was increased for SMF, although the slope halved relative to the rating for the original CGs. (c) A primary cue for viscosity was local image speed — an SMF looked less viscous just by increasing local speeds. (d) From the original SMFs containing spatiotemporal variations of local motion vectors, we made SMFs containing only spatial variations, and those containing only temporal variations. Liquidness rating for the spatial SMF was comparable to that for spatiotemporal SMF, but the rating for temporal SMF was significantly lower. Viscosity rating followed the simulated viscosity for spatial SMFs, but not for temporal SMFs. These results indicates that the human visual system can perceive the material property of moving fluid from optical flow information alone, and that the critical information resides in the local flow speed and spatial pattern of the optical flow field.
Meeting abstract presented at VSS 2013