A Modular GPU-based Direct Volume Renderer for Visualising Scalar and Multi-dimensional Data

Direct Volume Rendering (DVR) is a technique for displaying volumetric data sets without generating intermediate representations. Because of the increasing capabilities of computer hardware and consumer graphics accelerators, DVR can now be performed interactively and is becoming more and more popular. Originally, DVR has only been applied to scalar data sets, such as Computed Tomography scans. However, the dramatic advances in biomedical imaging and other research areas like meteorology and thermodynamics have resulted in ever increasingly complex volumetric data sets. Novel DVR algorithms are therefore required to deal with non-scalar and multivariate data at interactive frame rates. The ability to combine several data sets is also demanded, since insights can often be gained by comparing the difference or correlation between data sets. Current research in DVR mainly focuses on either improving the efficiency of rendering algorithms or the visual appearance of the renderings. Even though both aspects are of great importance, the efforts often lead to very specialised solutions that are only applicable in a certain domain. So far, no tool is available that works completely hardware-accelerated and, at the same time, supports the interactive exploration of arbitrary data as well as the easy integration of new data types and formats. In this report, we present solutions to develop a flexible and modular direct volume rendering framework which enables the user to interactively explore high-dimensional and multiple data sets on programmable consumer graphics cards. We discuss how the framework can be used to incorporate the latest specialised direct volume rendering algorithms and how it can be adapted on the fly to change a visualisation or to visualise new variables, such as the difference between two scalar fields, in order to gain insight into the given data. The capabilities of the framework are demonstrated by two simple case studies and the efficiency and effectiveness of the framework is evaluated.