Packing Vertex Data into Hardware-Decompressible Textures.

Area of Science:

• Computer Graphics
• Hardware Acceleration
• Data Compression

Background:

• Increasing scene complexity in 3D rendering strains graphics hardware memory resources.
• Vertex data compression is essential, requiring efficient decompression during rendering.

Purpose of the Study:

• To develop a novel method for vertex data decompression utilizing existing graphics processing unit (GPU) texture compression circuits.
• To optimize vertex data packing and permutation to minimize compression error and maintain data coherence.
• To enhance visual quality through vertex clustering for reduced quantization dynamic range.

Main Methods:

• Exploiting built-in texture compression hardware for real-time, random-order decompression of vertex data.
• Proposing an optimization approach for vertex data permutation to minimize compression error.
• Implementing vertex clustering to reduce data dynamic range during quantization.

Main Results:

• Achieved fast and high-quality vertex data decompression for real-time rendering.
• Demonstrated effectiveness across various 3D model vertex data.
• Resulted in a minimized memory footprint and high frame rates.

Conclusions:

• The proposed method efficiently leverages GPU texture compression for vertex data decompression.
• Optimization techniques ensure high visual quality and performance for real-time 3D graphics.
• This approach offers significant advantages in memory efficiency and rendering speed.