Relationship between scene complexity and perceptual performance for computer graphics simulations

Abstract An important goal for the design of visual displays is to determine how much realism or scene complexity to include in a computer simulation to reach a given level of performance. This is an important task since the present trend in computer graphics is to include the highest level of realism or scene complexity possible in a simulation. However, it may not be necessary to always include the highest level of realism or complexity to reach an acceptable level of performance. In fact, needless degrees of realism, and thus computational resources, may be wasted in the quest for ‘photographic’ realism. To study the relationship between scene complexity and human performance, three different simulations of scene complexity were modeled for a final approach task. The subject's task were to estimate two aspects of situation awareness, perceived altitude and aimpoint, during a simulated final approach at nine unique distances to threshold. The three levels of scene complexity included a homogeneous Lambertian shaded flat surface, farmlands, and farmlands with hills. The results indicated that increasing the level of scene complexity lead to better performance in judging both altitude and aimpoint during the simulated final approach. The relationship between scene complexity and perceptual performance for computer graphics simulations are discussed.