Position-based Torso Crowds
Abstract
We propose a novel crowd simulation method that represents crowd agents as 2D capsules rather than traditional discs. While discs are commonly used due to their simplicity, a disc does not accurately capture profile and orientation of crowd agents, and therefore misses intrinsic dynamic details of crowds in various settings. Leveraging position-based dynamics (PBD), we simulate crowds of capsule-shaped agents in multiple settings. We propose innovative constraints for multi-agent navigation with capsules, enhancing the fidelity and applicability of crowd simulations in diverse environments and scenarios.