Volume Rendering of Large Datasets

Many a small thing has been made large by the right kind of advertising.

Mark Twain

This chapter describes the fundamental components of our volume visualization system. First, we define a set of features that will be supported and derive the implications of these requirements. We then give an overview of the volume rendering pipeline. Next, the memory storage and access scheme that is used, is explained. We then go on to examine our approaches for parallelization and reconstruction. Finally, we show several optimizations to increase the performance of the algorithm and present a technique for increasing interactivity.

Subsections

• Introduction
  • System Requirements
    • Medical Datasets
    • High Quality
    • Interactive Viewing
    • Interactive Classification
    • Low Memory Consumption
    • Parallelization Support
    • Pure Software
  • Implications
  
  
• The Raycasting Pipeline
  • Reconstruction
    • Function Reconstruction
    • Gradient Reconstruction
  • Classification
  • Shading
  • Compositing
  
  
• Memory Management for Large Datasets
  • Bricking
  • Addressing
  • Traversal
  
  
• Parallelization Strategies for Commodity Hardware
  • Symmetric Multiprocessing
  • Simultaneous Multithreading
  
  
• Memory Efficient Acceleration Data Structures
  • Gradient Cache
  • Entry Point Buffer
  • Cell Invisibility Cache
  • Load Balancing
  
  
• Maintaining Interactivity