Generating Cultural Characters based on Hofstede Dimensions

The virtual humans behaviors can be endowed with different levels of intelligence and their action/motion can present more or less realism. In this scope, it is still a challenge to create, in an automatic and easy way, virtual humans that move and behave in a way that seems natural. Cultural aspects and population differences can produce deviations in speed, density and flow of the crowd. These aspects can be observed in videos of different Countries and certainly it can produce more realistic agents, if incorporated to simulation models. This paper presents a methodology to generate virtual humans which trajectories are based on real life video sequences, and in addition we considered Hofstede Cultural Dimensions (HCD) to map cultural differences. The method is tested in a set of real environments, but can be scaled for any environment that have video sequences and tracked human trajectories for training..