Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

3D Gaussian Particle Approximation of VDB Datasets: A Study for Scientific Visualization

Isha Sharma, Dieter Schmalstieg

TL;DR

This work investigates leveraging OpenVDB for SciVis by deriving a multi-LOD representation of voxel regions as 3D Gaussian particles and rendering them with GPU-accelerated ray marching via OptiX. The authors detail a data-generation workflow that maps leaf and grid-tree structures to non-overlapping Gaussians, implemented across low, medium, and high LODs, and they present a GPU-based renderer using GAS/IAS with per-Gaussian AABBs. Results on OpenVDB datasets show real-time performance (over 150 FPS) and favorable memory footprints at varying LODs, highlighting potential benefits for large sparse datasets and level-set representations in SciVis. The work also discusses limitations (non-overlapping Gaussians, need for advanced blending) and outlines concrete directions for enhancements, including overlapping Gaussians, smarter clustering, and improved volume compositing to enable smoother visualizations for scientific datasets.

Abstract

The complexity and scale of Volumetric and Simulation datasets for Scientific Visualization(SciVis) continue to grow. And the approaches and advantages of memory-efficient data formats and storage techniques for such datasets vary. OpenVDB library and its VDB data format excels in memory efficiency through its hierarchical and dynamic tree structure, with active and inactive sub-trees for data storage. It is heavily used in current production renderers for both animation and rendering stages in VFX pipelines and photorealistic rendering of volumes and fluids. However, it still remains to be fully leveraged in SciVis where domains dealing with sparse scalar fields like porous media, time varying volumes such as tornado and weather simulation or high resolution simulation of Computational Fluid Dynamics present ample number of large challenging data sets. The goal of this paper hence is not only to explore the use of OpenVDB in SciVis but also to explore a level of detail(LOD) technique using 3D Gaussian particles approximating voxel regions. For rendering, we utilize NVIDIA OptiX library for ray marching through the Gaussians particles. Data modeling using 3D Gaussians has been very popular lately due to success in stereoscopic image to 3D scene conversion using Gaussian Splatting and Gaussian approximation and mixture models aren't entirely new in SciVis as well. Our work explores the integration with rendering software libraries like OpenVDB and OptiX to take advantage of their built-in memory compaction and hardware acceleration features, while also leveraging the performance capabilities of modern GPUs. Thus, we present a SciVis rendering approach that uses 3D Gaussians at varying LOD in a lossy scheme derived from VDB datasets, rather than focusing on photorealistic volume rendering.

3D Gaussian Particle Approximation of VDB Datasets: A Study for Scientific Visualization

TL;DR

This work investigates leveraging OpenVDB for SciVis by deriving a multi-LOD representation of voxel regions as 3D Gaussian particles and rendering them with GPU-accelerated ray marching via OptiX. The authors detail a data-generation workflow that maps leaf and grid-tree structures to non-overlapping Gaussians, implemented across low, medium, and high LODs, and they present a GPU-based renderer using GAS/IAS with per-Gaussian AABBs. Results on OpenVDB datasets show real-time performance (over 150 FPS) and favorable memory footprints at varying LODs, highlighting potential benefits for large sparse datasets and level-set representations in SciVis. The work also discusses limitations (non-overlapping Gaussians, need for advanced blending) and outlines concrete directions for enhancements, including overlapping Gaussians, smarter clustering, and improved volume compositing to enable smoother visualizations for scientific datasets.

Abstract

The complexity and scale of Volumetric and Simulation datasets for Scientific Visualization(SciVis) continue to grow. And the approaches and advantages of memory-efficient data formats and storage techniques for such datasets vary. OpenVDB library and its VDB data format excels in memory efficiency through its hierarchical and dynamic tree structure, with active and inactive sub-trees for data storage. It is heavily used in current production renderers for both animation and rendering stages in VFX pipelines and photorealistic rendering of volumes and fluids. However, it still remains to be fully leveraged in SciVis where domains dealing with sparse scalar fields like porous media, time varying volumes such as tornado and weather simulation or high resolution simulation of Computational Fluid Dynamics present ample number of large challenging data sets. The goal of this paper hence is not only to explore the use of OpenVDB in SciVis but also to explore a level of detail(LOD) technique using 3D Gaussian particles approximating voxel regions. For rendering, we utilize NVIDIA OptiX library for ray marching through the Gaussians particles. Data modeling using 3D Gaussians has been very popular lately due to success in stereoscopic image to 3D scene conversion using Gaussian Splatting and Gaussian approximation and mixture models aren't entirely new in SciVis as well. Our work explores the integration with rendering software libraries like OpenVDB and OptiX to take advantage of their built-in memory compaction and hardware acceleration features, while also leveraging the performance capabilities of modern GPUs. Thus, we present a SciVis rendering approach that uses 3D Gaussians at varying LOD in a lossy scheme derived from VDB datasets, rather than focusing on photorealistic volume rendering.

Paper Structure

This paper contains 18 sections, 1 equation, 4 figures, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. Goal
  3. Previous works
  4. Data Generation Model
  5. Grid Layout
  6. 3D Volumetric Gaussians generation
  7. Leaf nodes Conversion
  8. LOD implementation
  9. Gaussians Renderer
  10. Acceleration Data Structures
  11. Ray Tracing Pipeline
  12. Intersection
  13. Rendering Loop
  14. Transfer functions
  15. Combining Volumes and Level-sets
  16. ...and 3 more sections

Figures (4)

  • Figure 1: 3D Gaussians approximating VDB leaf nodes in the bunny_cloud dataset.
  • Figure 2: OpenVDB datasets fire(left) and explosion(right) in two different transfer functions. At the top row we have jet and in the middle viridis. At the bottom row we have photorealistic rendering with NanoVDB HDDA transmittance for comparison
  • Figure 3: Combined surface and volume modeling with 3D Gaussians
  • Figure 4: Top to Bottom: bunny_cloud, utahteapot, venusstatue and explosion in three LOD setting i.e., left-to-right: High, Medium and Low