Available under a Creative Commons Attribution Non-Commercial Share Alike 4.0 International Licence
1.2 COMPUTER AND INFORMATION SCIENCE
Virtual humans are often endowed with human-like characteristics to make them more appealing and engaging. Motion capture is a reliable way to represent natural motion on such characters, thereby allowing a wide range of animations to be automatically created and replicated. However, interpersonal differences in actors’ performances can be subtle and complex, yet have a strong effect on the human observer. Such effects can be very difficult to express quantitatively or indeed even qualitatively. We investigate two subjective human motion characteristics: attractiveness and distinctiveness. We conduct a perceptual experiment, where participants’ eye movements are tracked while they rate the motions of a range of actors. We found that participants fixate mostly on the torso, regardless of gait and actor sex, and very little on the limbs. However, they self-reported that they used hands, elbows and feet in their judgments, indicating a holistic approach to the problem.
Ennis, C., Hoyet, L., & O'Sullivan, C. (2015). Eye-tracktive: measuring attention to body parts when judging human emotions. Eurographics 2015 (Short papers), pp37-40. doi:10.2312/egsh.20151009