UnrealVis: A Testing Laboratory of Optimization Techniques in Unreal Engine for Scientific Visualization
Researchers built a lab in Unreal Engine to test 22 rendering optimizations for massive scientific datasets.
A team from Sapienza University of Rome has developed UnrealVis, a novel framework that turns Unreal Engine 5 into a configurable laboratory for optimizing the visualization of massive 3D scientific datasets. The researchers systematically reviewed 55 academic papers to create a taxonomy of 22 distinct optimization techniques, which they implemented across six families within the engine. These include leveraging cutting-edge subsystems like Nanite for virtualized geometry, Level of Detail (LOD) schemes, and various culling methods. The core value of UnrealVis is its intuitive workflow that lets users—often domain scientists without deep graphics programming expertise—experiment with different optimization combinations in real-time.
UnrealVis features a live telemetry dashboard and A/B comparison tools, enabling both local and global performance analysis. This allows researchers to see precisely how each technique impacts frame rates and memory usage while exploring their data. The system was validated through case studies visualizing complex biological structures like ribosomes and volumetric flow fields, followed by an expert evaluation. The result is a practical tool that helps scientists select the right blend of optimizations to hit specific performance targets, such as maintaining 60 FPS during interactive exploration, while crucially preserving the structural fidelity necessary for accurate analysis. The project is openly available, providing a much-needed bridge between high-performance game engine technology and the demanding world of scientific visualization.
- Built on Unreal Engine 5, implementing 22 optimization techniques (like Nanite & LOD) from a review of 55 papers.
- Features live telemetry and A/B testing for real-time performance analysis during dataset exploration.
- Validated with case studies on ribosomal structures and flow fields, helping scientists balance speed and visual accuracy.
Why It Matters
Democratizes high-performance 3D visualization for scientists, letting them achieve interactive frame rates on complex datasets without expert graphics coding.