UnrealVis: A Testing Laboratory of Optimization Techniques in Unreal Engine for Scientific Visualization

Abstract

Visualizing large 3D scientific datasets requires balancing performance and fidelity, but traditional tools often demand excessive technical expertise. We introduce UnrealVis, an Unreal Engine optimization laboratory for configuring and evaluating rendering techniques during interactive exploration. Following a review of 55 papers, we established a taxonomy of 22 optimization techniques across six families, implementing them through engine subsystems such as Nanite, Level of Detail(LOD) schemes, and culling. The system features an intuitive workflow with live telemetry and A/B comparisons for local and global performance analysis. Validated through case studies of ribosomal structures and volumetric flow fields, along with an expert evaluation, UnrealVis facilitates the selection of optimization combinations that meet performance goals while preserving structural fidelity. UnrealVis is available at https://github.com/XAIber-lab/UnrealVis

Figures (6)

• Figure 1: The UnrealVis optimization taxonomy: categories and specific techniques synthesized from the literature review and engine-native capabilities, divided into visualization-centric (top) and simulation-centric methods (bottom). The color-coding assigned to the six primary families is mirrored in the UnrealVis user interface. The color legend reports how many times the optimization techniques appear in the literature.
• Figure 2: Technical schematic of the UnrealVis data ingestion pipeline, highlighting the dual-path architecture: the asynchronous C++ module for PDB retrieval and the Python-integrated scripting layer for custom dataset processing and OpenEXR multiphysics transcoding.
• Figure 3: An example workflow for starting a simulation in UnrealVis, showing a huge bacterial 70S ribosome. The user is first greeted by a welcome screen, displaying a human 17$\beta$-hydroxysteroid dehydrogenase type 1 (oxido-reductase enzyme, $\sim$35 kDa) complexed with NADP$^+$ cofactor and equilin inhibitor pdb_1equSawicki1999 by default. A menu shows the different views selection on the left, while a small panel on the right briefly describes the loaded dataset. To begin the workflow, the user selects the dataset from the list (A), then chooses the desired optimizations from the appropriate menu (B). After renaming the simulation and inputting its description (C), the simulation can start. The user can freely move within the 3D dataset using the mouse and keyboard, while performance metrics are captured in real time (D). Then, the user can analyze performance statistics relative to simulations by directly comparing two simulations over time (E) or by inspecting performance using small multiples when simulations are many (F).
• Figure 4: 3SYJ adhesin exploration. The two-way whisker isolates the oligomerization interface (black), balancing high-fidelity rendering with global navigation performance.
• Figure 5: BLASTNet 2.0 volumetric exploration. The main view tracks real-time telemetry and optimization settings. Bottom insets provide multi-angle perspectives: the left inset highlights internal structures using a cyan-to-magenta gradient, while the right inset captures the high-vorticity red core from a wide-angle frontal view.