Graphics: Visual and Interactive Computing

• VIC  | 
• Volume Graphics  | 
• Data Visualization  | 
• Interactive  | 
• Intelligent  | 
• Members

Volume Graphics

Academics: M. Chen, M. W. Jones, B. Mora

Collaborators: Rutgers (USA), Purdue (USA), Bath (UK), AAS (Austria), DTU (Denmark), Stuttgart (Germany), Swansea Theory Group (UK)

The group has established its leading position internationally in the field of volume graphics, which was evolved from volume visualization, and aims at developing volumetric techniques into a general-purpose graphics technology. The group has enabled volume graphics to match, and in some aspects supersede, surface graphics. New methods were developed for modelling, rendering, manipulating, deforming and animating graphics models in a true 3D manner. In particular, the group members have made a number of significant contributions to the fundamentals of volume graphics.

The algebraic framework of constructive volume geometry (CVG) was first developed at Swansea, and is a major generalization of CSG. It operates on solid objects as well as amorphous phenomena, and the interior as well exterior of objects. It provides vlib (an open source software API for volume graphics) with a constructive modelling framework for volume scene graphs. Recently new concepts were introduced to CVG, enabling the direct modelling and rendering of deformation and point-based objects.

3D distance fields (DFs) provide a cost-effective means for modelling surface objects in volume graphics. Following its first introduction by the group for volume reconstruction from contour stacks, a series of new and efficient techniques were developed for computing accurate DFs in voxelization, compressing DFs and employing DFs in hypertexture and deformation.

Dr. Mora, who joined the group two years ago, has a wealth of research experience in volume rendering. In particular, a new object-order algorithm was developed for fast volume ray-casting, which was commonly considered as an image-order technique. An algorithmic framework was proposed and studied for a class of order-independent volume rendering algorithms which were typically suitable for interactive volume graphics. A recent fundamental study on the maximum intensity projection (MIP) algorithm showed that MIP could be performed in O(1) time per pixel by exploiting spatial coherence effectively. This is a significant finding that provides an optimal lower bound for image-space direct volume rendering.

Volume deformation and animation remains one of the most challenging subjects in volume graphics. The group developed a number of new techniques, many in collaboration with Rugters (USA) and Bath (UK), for modelling and animating a variety of effects such as free-form deformation, sweeping, splitting, melting and explosion. The concept of spatial transfer function was formulated for deforming volume objects as part of volume scene graphs. The technique of volume wires was introduced for modelling skeleton-based deformation using volume sweeping. More recently, the concept of displacement mapping was generalized to accommodate complex volumetric displacements by lifting the constraints on displacement continuity, direction and magnitude.

There are a great number of other contributions by the group members, including non-photorealistic volume rendering, isosurface visualization with parametric cubes, parallel volume rendering, shadows and refraction in volume graphics, volume morphing, and acceleration techniques for volume rendering. One of the latest developments is the introduction of a new spectral volume rendering integral based on the Kubelka and Munk theory instead of the Lambert-Bouguer law, allowing for more accurate modelling of light transport in volumetric media. In collaboration with Stuttgart (Germany), the new integral was successfully realized on GPU for real-time volume graphics.