QuadStack: An Efficient Representation and Direct Rendering of Layered Datasets.

Area of Science:

• Computer Science
• Data Compression
• Scientific Visualization

Background:

• Volumetric data is prevalent in scientific and engineering fields.
• Existing compression and rendering methods struggle with large, layered datasets.
• Data redundancy in layered structures is often unexploited.

Purpose of the Study:

• To introduce QuadStack, a novel algorithm for volumetric data compression and direct rendering.
• To exploit data redundancy in layered scientific datasets for improved efficiency.
• To enable fast data access and high-quality rendering with reduced memory footprint.

Main Methods:

• QuadStack compresses volumetric data into vertical stacks, then into a quadtree.
• Layered structures are identified and represented at internal quadtree nodes.
• Associated data (color, material, density) and shape are decoupled and encoded independently.
• A novel algorithm for value retrieval with logarithmic complexity is introduced.

Main Results:

• Achieved compression rates from 4x to 54x the original memory footprint.
• Demonstrated logarithmic complexity for data value retrieval.
• GPU implementation achieved speeds comparable to state-of-the-art algorithms (18-79 MRays/s).
• Maintained a significantly smaller memory footprint compared to existing methods.

Conclusions:

• QuadStack offers efficient volumetric data compression and direct rendering.
• The algorithm's performance is suitable for large-scale scientific and engineering applications.
• QuadStack provides a balance of high compression, fast access, and competitive rendering speeds.