Using the Quest to Explore Large Scientific Data in Virtual Reality
Click the Poster to View Full Screen, Right click to save image
Idunnuoluwa Adeniji
College:
The Dorothy and George Hennings College of Science, Mathematics, and Technology
Major:
Computational Science & Engineering
Faculty Research Advisor(s):
David Joiner
Abstract:
Title: Using the Quest to Explore Large Scientific Data in Virtual Reality
Introduction:
Traditional scientific data exploration relies on 2D and 3D visualization tools. However, a transformation is occurring with lower cost virtual reality (VR) hardware. This study showcases a data pipeline developed from ParaView to Unity Game Engine and the Oculus Quest 2 headset to explore this transition.
Objectives:
This project applied VR to the identification of patterns and clusters within point cloud data, in particular data generated by a Hardware/Hybrid Accelerated Cosmology Code (HACC) simulation. We enabled effective user interaction by integrating VR into the broader field of large data exploration, which includes features like data interaction, manipulation, and in-depth analysis. We implemented functions to pinpoint the position and value of dark matter halos simply by pointing at them. These custom interactions are crucial because they extend beyond the capabilities of ParaView's VR mode.
Methodology:
We use a custom script to import a sample of approximately 4 million particles (out of up to 2 trillion simulated), generated by a HACC simulation. Additional halo and iso surface data is also introduced into the VR environment. The point cloud data is rendered as a point topology mesh using a custom geometry shader, the halos are represented by spherical game objects, and the isos are imported as it is rendered in ParaView. These sphere objects offer interactive functionality, allowing users to interact with them and retrieve specific details.
Results:
Through VR immersion, this study significantly amplifies the identification of patterns and clusters when compared to conventional methods. In our HACC visualization, we implemented the ability to highlight each halo as part of the visualization user interface. In addition to these features, we achieved an average PC frame rate of approximately 1012.3 frames per second (FPS) when visualizing 10,000 particles on a Alienware Aurora R15 with a GeForce 3090, and for 4 million particles, we attained an average PC frame rate of approximately 235.9 FPS.