Representational validity of landscape visualizations: The effects of graphical realism on perceived scenic beauty of forest vistas

Photographs have long been used to represent environmental conditions in the context of landscape quality assessments and environmental perception research. Representational options have been significantly expanded by applications of computer modeling and computer graphic technologies that can provide precise visualizations based on inventoried or model-projected biophysical data. Final graphic displays from computer visualization systems can range from very abstract 'wire-frame' models to high resolution, photorealistic video images. An important assumption underlying the use of both photographic and computer rendered visualizations is that human viewers' responses to these representations provide valid indications of perceptions and judgments made in response to direct experience with the landscape conditions nominally represented. In this study the same set of forest landscape scenes was represented by visualizations rendered at four different levels of realism-abstraction. Each representation was shown to separate groups of observers who rated the perceived scenic beauty of the common set of forest landscape scenes. Correlations between the ratings of the same scenes in the different visualization conditions were very low, raising important questions about the representational validity of computer-generated landscape visualizations.