2D Gaussian Splatting for Geometrically Accurate Radiance Fields
arXiv:2403.17888 · doi:10.1145/3641519.3657428
Abstract
3D Gaussian Splatting (3DGS) has recently revolutionized radiance field reconstruction, achieving high quality novel view synthesis and fast rendering speed without baking. However, 3DGS fails to accurately represent surfaces due to the multi-view inconsistent nature of 3D Gaussians. We present 2D Gaussian Splatting (2DGS), a novel approach to model and reconstruct geometrically accurate radiance fields from multi-view images. Our key idea is to collapse the 3D volume into a set of 2D oriented planar Gaussian disks. Unlike 3D Gaussians, 2D Gaussians provide view-consistent geometry while modeling surfaces intrinsically. To accurately recover thin surfaces and achieve stable optimization, we introduce a perspective-correct 2D splatting process utilizing ray-splat intersection and rasterization. Additionally, we incorporate depth distortion and normal consistency terms to further enhance the quality of the reconstructions. We demonstrate that our differentiable renderer allows for noise-free and detailed geometry reconstruction while maintaining competitive appearance quality, fast training speed, and real-time rendering.
13 pages, 12 figures. Corrected Eq.7
References in corpus (4)
Cited by in corpus (34)
- 3D Gaussian as a New Era: A Survey
- PGSR: Planar-based Gaussian Splatting for Efficient and High-Fidelity Surface Reconstruction
- VR-Splatting: Foveated Radiance Field Rendering via 3D Gaussian Splatting and Neural Points
- ActiveSplat: High-Fidelity Scene Reconstruction through Active Gaussian Splatting
- FeatureGS: Eigenvalue-Feature Optimization in 3D Gaussian Splatting for Geometrically Accurate and Artifact-Reduced Reconstruction
- Explicit Correspondence Matching for Generalizable Neural Radiance Fields
- GSORB-SLAM: Gaussian Splatting SLAM benefits from ORB features and Transmittance information
- Image-GS: Content-Adaptive Image Representation via 2D Gaussians
- HI-SLAM2: Geometry-Aware Gaussian SLAM for Fast Monocular Scene Reconstruction
- VR-Robo: A Real-to-Sim-to-Real Framework for Visual Robot Navigation and Locomotion
- Robust LiDAR-Camera Calibration with 2D Gaussian Splatting
- GS-ROR: Bidirectional-guided 3DGS and SDF for Reflective Object Relighting and Reconstruction
- When Gaussian Meets Surfel: Ultra-fast High-fidelity Radiance Field Rendering
- As-Rigid-As-Possible Deformation of Gaussian Radiance Fields
- SplatMAP: Online Dense Monocular SLAM with 3D Gaussian Splatting
- Radiance Surfaces: Optimizing Surface Representations with a 5D Radiance Field Loss
- Complex-Valued Holographic Radiance Fields
- Scalable and High-Quality Neural Implicit Representation for 3D Reconstruction
- DynamicGSG: Dynamic 3D Gaussian Scene Graphs for Environment Adaptation
- Cutting-edge 3D reconstruction solutions for underwater coral reef images: A review and comparison
- 2DGH: 2D Gaussian-Hermite Splatting for High-quality Rendering and Better Geometry Features
- GS-RoadPatching: Inpainting Gaussians via 3D Searching and Placing for Driving Scenes
- GS-ProCams: Gaussian Splatting-based Projector-Camera Systems
- High-quality Animatable Eyelid Shapes from Lightweight Captures
- EfficientHuman: Efficient Training and Reconstruction of Moving Human using Articulated 2D Gaussian
- DeMapGS: Simultaneous Mesh Deformation and Surface Attribute Mapping via Gaussian Splatting
- High-Quality Spatial Reconstruction and Orthoimage Generation Using Efficient 2D Gaussian Splatting
- Animus3D: Text-driven 3D Animation via Motion Score Distillation
- Points as Tori: Fast Pointwise Signed Distance for Point Clouds
- Vector Scaffolding: Inter-Scale Orchestration for Differentiable Image Vectorization
- DR-GS: Physically-Based Deformable and Relightable 2D Gaussians
- LN3DIFF++: Scalable Latent Neural Fields Diffusion for Speedy 3D Generation
- Gaussian Building Mesh (GBM): Extract a Building's 3D Mesh with Google Earth and Gaussian Splatting
- From Volume Rendering to 3D Gaussian Splatting: Theory and Applications