Proceedings of the Annual Meeting of the Cognitive Science Society

For creating photorealistic images, Computer Graphics re-searchers introduced Bidirectional Texture Functions (BTFs),which use view- and illumination-dependent textures for ren-dering. BTFs require massive storage, and several proposalswere made on how to compress them, but very few take intoaccount human perception. We present and discuss an exper-imental study on how decreasing the texture resolution influ-ences perceived quality of the rendered images. In a visualcomparison task, observer quality judgments and gaze datawere collected and analysed to determine the optimal down-sampling of BTF data without significant loss of their per-ceived visual quality.