Interpolating and Downsampling RGBA Volume Data

Martin Kraus and Kai Bürger
Computer Graphics and Visualization Group, Technische Universität München, Germany

Background

For about twenty years the standard color sampling method for RGBA volume data has been the interpolation of opacity-weighted colors. In this work, we discuss the underlying approximations and derive an improved sampling method, which avoids some of these approximations. With the help of several examples and experiments we demonstrate that our new technique is in fact preferable for direct volume rendering of RGBA data. Moreover, we show that the same improvement can be applied to downsampling filters. Finally, we propose a procedure to design improved downsampling filters for the generation of pyramid RGBA volume data such as mipmap volume textures.