Visuals are not what they look

Abstract

When developing virtual environments (VEs), most effort goes into developing the visuals. For many, the ideal is to create virtual worlds of photo-realistic quality or otherwise being of high fidelity. The purpose is to make the VE seem real to the user. This paper takes a closer look at subjects’ ratings of the visuals, and of the extent to which the VE feels real to the subjects, in the context of an experiment on audio in which subjects were to perform two search tasks: the first in an ordinary, textured house; the second in a bare structure consisting almost exclusively of the barren, white walls. Audio was never relevant to the search task in the first experiment, while in the second experiment it was relevant to the search task for half of the subjects. Subjects for whom audio was irrelevant to both their search tasks rated their visual involvement as large in the barren VE as in the higher quality one. However, subjects for which audio was relevant to their search task in the second experiment saw their visual involvement plummet, while their auditory involvement surged. Finally, the extent to which the VE felt real to the subjects did not correlate with their visual involvement but instead showed a strong correlation with the extent to which the interaction felt natural.