Abstract:Neural Radiance Fields (NeRFs) have demonstrated remarkable potential in capturing complex 3D scenes with high fidelity. However, one persistent challenge that hinders the widespread adoption of NeRFs is the computational bottleneck due to the volumetric rendering. On the other hand, 3D Gaussian splatting (3DGS) has recently emerged as an alternative representation that leverages a 3D Gaussisan-based representation and adopts the rasterization pipeline to render the images rather than volumetric rendering, achieving very fast rendering speed and promising image quality. However, a significant drawback arises as 3DGS entails a substantial number of 3D Gaussians to maintain the high fidelity of the rendered images, which requires a large amount of memory and storage. To address this critical issue, we place a specific emphasis on two key objectives: reducing the number of Gaussian points without sacrificing performance and compressing the Gaussian attributes, such as view-dependent color and covariance. To this end, we propose a learnable mask strategy that significantly reduces the number of Gaussians while preserving high performance. In addition, we propose a compact but effective representation of view-dependent color by employing a grid-based neural field rather than relying on spherical harmonics. Finally, we learn codebooks to compactly represent the geometric attributes of Gaussian by vector quantization. With model compression techniques such as quantization and entropy coding, we consistently show over 25$\times$ reduced storage and enhanced rendering speed, while maintaining the quality of the scene representation, compared to 3DGS. Our work provides a comprehensive framework for 3D scene representation, achieving high performance, fast training, compactness, and real-time rendering. Our project page is available at <a class="link-external link-https" href="https://maincold2.github.io/c3dgs/" rel="external noopener nofollow">this https URL</a>.

Fast annual daylighting simulation and high dynamic range image processing using NumPy

Deep Neural Network Approach for Annual Luminance Simulations

Fast, Realistic Lighting and Material Design Using Nonlinear Cut Approximation

Lighting Every Darkness with 3DGS: Fast Training and Real-Time Rendering for HDR View Synthesis

A Matrix Sampling-And-Recovery Approach for Many-Lights Rendering

Adaptive Matrix Column Sampling and Completion for Rendering Participating Media.

Constant-Cost Spatio-Angular Prefiltering of Glinty Appearance Using Tensor Decomposition

EAGLES: Efficient Accelerated 3D Gaussians with Lightweight EncodingS

Efficient matrix algebra encoding for urban solar irradiation simulation: fine-grid ground-level estimation with vector data

Parallel Viewshed Analysis on GPU Using CUDA

LightGaussian: Unbounded 3D Gaussian Compression with 15x Reduction and 200+ FPS

Research on 3D Visualization of Drone Scenes Based on Neural Radiance Fields

NARVis: Neural Accelerated Rendering for Real-Time Scientific Point Cloud Visualization

EffLiFe: Efficient Light Field Generation via Hierarchical Sparse Gradient Descent

Depth Guided Fast Rendering Neural Radiance Fields

Compact 3D Gaussian Representation for Radiance Field

HybridNeRF: Efficient Neural Rendering via Adaptive Volumetric Surfaces

Fast High Dynamic Range Radiance Fields for Dynamic Scenes

A Multilayer Perceptron-based Fast Sunlight Assessment for the Conceptual Design of Residential Neighborhoods under Chinese Policy

PyGS: Large-scale Scene Representation with Pyramidal 3D Gaussian Splatting

DNGaussian: Optimizing Sparse-View 3D Gaussian Radiance Fields with Global-Local Depth Normalization