Efficient Image-Based Bidirectional Texture Function Model

Recent advances in computer hardware and virtual modelling allow to respect view and illumination dependencies of natural surface materials. The corresponding texture representation in the form of Bidirectional Texture Function (BTF) enables significant improvements of virtual models realism at the expense of immense increase of material sample data space. Thus introduction of some fast compression, modelling and rendering method for BTF data is inevitable. In this paper we introduce a generalisation of our polynomial extension of the Lafortune model computed for every original BTF measurement pixel allowing seamless BTF texture enlargement. This nonlinear reflectance model is further extended using parameters clustering technique to achieve higher compression ratio. The presented method offers BTF modelling in excellent visual quality as was tested on variety of BTF measurements. The method gives BTF compression ratio ∼1:200 as well as fast graphics hardware implementation.