The approach to volume rendering which is presented in this work aims on providing interactive frame rates even on low-end hardware. To achieve this goal, a set of optimizations and efficient techniques are combined for data preprocessing and rendering. The algorithms are based on an interpretation of the volume as a set of voxels defined on a Cartesian grid. Key features of the concept are
Data sets as dealt with in the context of this work are considered to be composed of objects, i.e., semantically distinct sub-sets of the data. Therefore, each voxel of the volume is considered to belong to an object, i.e., a spatial structure within the data. The representation of extracted voxels allows a very flexible handling and assignment of visualization and rendering parameters, on a per-object basis. This feature can be exploited to subdivide the volume in a way which corresponds to the internal structure of the contained objects. Each object can be rendered using the color, opacity, shading, and compositing method which is best-suited to obtain the desired visualization goal. The flexibility of visualization-parameter tuning and especially the ability to mix different shading models and compositing modes within a single visualization is not achievable with many other volume rendering approaches. Especially in combination with the interactivity of rendering, this poses a unique feature of our approach.