Octree rasterization: accelerating high-quality out-of-core GPU volume rendering.

Area of Science:

• Computer Graphics
• Scientific Visualization

Background:

• Large-scale volume data rendering presents significant computational challenges.
• Traditional GPU rendering algorithms struggle with efficiency for out-of-core datasets.

Purpose of the Study:

• To develop a novel GPU-based approach for high-quality volume rendering of large out-of-core data.
• To significantly reduce rendering time by optimizing ray traversal and empty-space culling.

Main Methods:

• Utilized an octree structure (of bricks), with bricks further subdivided into macrocells.
• Moved accelerating structure traversal off the GPU raycasting loop.
• Implemented an efficient empty-space culling method using rasterization of proxy geometry to generate per-pixel lists of penetrated bricks.

Main Results:

• Achieved two levels of empty-space skipping: brick level and macrocell level.
• Rendering speed was 2 to 4 times faster than state-of-the-art algorithms.
• Demonstrated effectiveness across a variety of datasets.

Conclusions:

• The proposed method offers substantial performance gains for rendering large out-of-core volume data.
• Optimized ray traversal and hierarchical data structures are key to efficient GPU volume rendering.