Impact of Anomalies in Simulated Crowds

Abstract: Virtual simulations of crowds are useful for many different fields of research and entertainment, such as simulations of evacuation scenarios or in video games that use large amounts of moving agents. This study examines if the noticeability of movement anomalies are dependent on their location on the screen and the density of the virtual crowd. The study also examines if movement anomalies change the perceived realism of the crowd simulation even if it is not consciously found. An user experiment with 12 participants was conducted, where the subjects watched 48 videos recorded from low and high density crowd simulation scenarios. An eye tracker gathered data from the participants while they watched the videos. Half of these videos had movement anomalies in specific locations. For each video, the observer filled in a part of a questionnaire to answer if they perceived the video to be realistic and if they saw a movement anomaly. The results showed a small correlation between anomalies in the center of the screen and an increase in noticeability. They also showed that the density of the crowd will have an impact on the detection rate of an anomaly. Finally, the results showed no indication that unnoticed anomalies would effect the subjects perceived realism.