Advances in 3D Gaussian Splatting

The field of 3D rendering is witnessing significant advancements with the development of 3D Gaussian Splatting techniques. Recent research has focused on improving the efficiency, fidelity, and versatility of these methods. Notably, innovative approaches have been proposed to enhance compression, initialization, and density control in 3D Gaussian Splatting, enabling high-quality real-time rendering and improved visual fidelity. Furthermore, efforts have been made to address aliasing artifacts and integrate traditional computer graphics techniques with Gaussian-splatted environments. These developments are pushing the boundaries of what is possible in real-time rendering and opening up new avenues for applications in various fields. Noteworthy papers include: Confident Splatting, which presents a novel lossy compression method based on learnable confidence scores. GDGS, which introduces a geometry-guided initialization and dynamic density control mechanism for improved rendering quality. LOD-GS, which proposes a Level-of-Detail-sensitive filtering framework for eliminating aliasing artifacts. Gbake, which enables realistic reflection mapping of traditional 3D meshes in Gaussian-splatted environments. Real-time Image-based Lighting of Glints, which provides an efficient approximation for image-based lighting of glints in real-time rendering applications.

Sources

Confident Splatting: Confidence-Based Compression of 3D Gaussian Splatting via Learnable Beta Distributions

GDGS: 3D Gaussian Splatting Via Geometry-Guided Initialization And Dynamic Density Control

LOD-GS: Level-of-Detail-Sensitive 3D Gaussian Splatting for Detail Conserved Anti-Aliasing

Gbake: Baking 3D Gaussian Splats into Reflection Probes

Real-time Image-based Lighting of Glints

Built with on top of