Reproduce “Super-Gaussian: Interactive Scene Editing for 3D Gaussian Splatting and NLI-Based Volume Visualization in Virtual Reality” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Suemin Jeon, Kaiyuan Tang, Chaoli Wang, Won-Ki Jeong
The Super-Gaussian framework enables interactive scene editing and NLI-driven volume visualization within virtual reality environments, utilizing editable 3D Gaussian Splatting and a structure-aware ROI selection mechanism. This system maintains real-time performance, near 120 FPS, while demonstrating superior selection accuracy and efficiency compared to existing methods.
Key insights
- Editable 3D Gaussian Splatting effectively provides real-time, high-frame-rate rendering for volumetric data in VR, outperforming traditional Direct Volume Rendering.
- Grouping Gaussian primitives into "Super-Gaussians" through geodesic SLIC clustering creates structurally coherent units, significantly improving the precision and efficiency of interactive ROI selection.
- Integrating Natural Language Interaction (NLI) with VR-based spatial selection enables intuitive, multimodal semantic labeling, querying, and manipulation of volumetric regions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.