next up previous contents
Next: Space-Efficient Object Representation for Up: Interactive Rendering Previous: Results and Discussion   Contents

Summary

The previous sections presented the application of the proposed techniques, i.e., the identification of relevant voxels and their extraction and storage in RenderLists, to various rendering techniques. Depending on the desired compositing technique, lighting model, and transfer function setup, different optimizations and voxel ordering schemes can be applied to optimize performance. The bit resolution used to store voxel coordinates (8-11 bit per coordinate) and quantized gradient vectors (12-16 bit) for shading also varies between the described techniques, to optimize performance for a given application.

In chapter 6 a unified implementation will be presented which combines the presented techniques within a single framework, using similar bit resolution and voxel ordering for all techniques to allow fast and flexible parameter modification and rendering mode changes. Although the unified implementation is achieved at the cost of slightly lower rendering performance than highly optimized special-purpose implementations, the rendering speed is still interactive even on standard desktop hardware.


Lukas Mroz, May 2001,
mailto:mroz@cg.tuwien.ac.at.