Técnicas de visão panorâmica e de percepção de densidade para modelos de simulação de multidão baseados em visão sintética.

Abstract

Crowd simulation is important in a wide variety of virtual reality applications, since the existence of avatars capable of behaving like a real crowd improves the user’s immersion sense. Recently, models to simulate crowds with characters that have visual perception, for being equipped with a synthetic vision system, have brought interesting results in terms of naturality in which the characters move themselves through the environment. In this work, two new functionalities for the virtual characters synthetic vision systems are proposed: the first is the substitution of the planar projection, originally used for the vision of the agents, for a technique that renders panoramic scenes using slices, to allow field of views with angles greater than 180º. The impact of this change is evaluated both on the performance of the implementation, but also how it changes the behavior of the agent. The second addition is the perception of density of crowds over the gradient-based model. A comparative study is performed over this model with the existing technique.