Shape Shifters: Does Body Shape Change the Perception of Small-Scale Crowd Motions?

TL;DR

This work investigates whether body-shape diversity modulates the perceptual detectability of motion clones in small-scale crowds of physics-based avatars. Using a Unity-based ragdoll system with a PD controller to follow reference motions, the authors created Baseline crowds with unique motions and shapes and Motion Clone crowds with varying levels of motion cloning and body-shape diversity, then conducted a side-by-side perception task across 22 participants. The results show that increasing motion variety reduces clone detectability, while body-shape diversity has no robust effect, suggesting motion cues dominate perceived crowd realism in this setting. The findings inform efficient crowd rendering by emphasizing motion variety over exhaustive appearance diversification for small crowds, with implications for real-time graphics and VR applications.

Abstract

The animation of realistic virtual avatars in crowd scenarios is an important element of immersive virtual environments. However, achieving this realism requires attention to multiple factors, such as their visual appearance and motion cues. We investigated how body shape diversity influences the perception of motion clones in virtual crowds. A physics-based model was used to simulate virtual avatars in a small-scale crowd of size twelve. Participants viewed side-by-side video clips of these virtual crowds: one featuring all unique motions (Baseline) and the other containing motion clones (i.e., the same motion used to animate two or more avatars in the crowd). We also varied the levels of body shape and motion diversity. Our findings revealed that body shape diversity did not influence participants' ratings of motion clone detection, and motion variety had a greater impact on their perception of the crowd. Further research is needed to investigate how other visual factors interact with motion in order to enhance the perception of virtual crowd realism.

Figures (5)

• Figure 1: Ragdoll representation of an avatar with rigid bodies. This diagram shows the structure of a ragdoll created using primitive-shaped rigid bodies and configurable joints.
• Figure 2: Stimulus Creation: Motion Clone (left): Featuring characters with two distinct motions, the first circled in green, and the second in red (out-of-step motions). Each character has one of six different body shapes. Baseline (right): All characters have unique body shapes and unique motions.
• Figure 3: Example Stimulus: Participants indicated their confidence level when selecting the side on which all motions were unique.
• Figure 4: Mean participant responses (with standard error bars) showing accuracy for identifying the gold standard crowd. Higher values indicate more accurate motion clone detection.
• Figure 5: Distribution of key factors considered by participants during experiment.