CLoD-GS: Continuous Level-of-Detail via 3D Gaussian Splatting

2025-10-10

AI Summary

Traditional discrete level-of-detail (DLoD) methods require storing multiple model variants, causing visual popping artifacts during transitions and incurring substantial memory overhead. To address this, we propose the first continuous level-of-detail (CLoD) framework for 3D Gaussian splatting. Our method introduces a learnable, distance-dependent opacity decay mechanism, virtual distance scaling, and a coarse-to-fine training strategy to enable view-distance-adaptive, artifact-free detail transitions within a single model. Additionally, we incorporate rendering-point regularization and dynamic sampling control to significantly reduce primitive count and GPU memory consumption while preserving rendering fidelity. Experiments demonstrate that CLoD completely eliminates flickering and geometric popping, enabling high-fidelity, scalable real-time rendering across diverse performance targets.

Technology Category

• Computer Vision: 3D Computer Vision
• Search and Optimization: Mixed Discrete/Continuous Search
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Security and Privacy: Data transparency and provenance

Abstract

Level of Detail (LoD) is a fundamental technique in real-time computer graphics for managing the rendering costs of complex scenes while preserving visual fidelity. Traditionally, LoD is implemented using discrete levels (DLoD), where multiple, distinct versions of a model are swapped out at different distances. This long-standing paradigm, however, suffers from two major drawbacks: it requires significant storage for multiple model copies and causes jarring visual ``popping" artifacts during transitions, degrading the user experience. We argue that the explicit, primitive-based nature of the emerging 3D Gaussian Splatting (3DGS) technique enables a more ideal paradigm: Continuous LoD (CLoD). A CLoD approach facilitates smooth, seamless quality scaling within a single, unified model, thereby circumventing the core problems of DLOD. To this end, we introduce CLoD-GS, a framework that integrates a continuous LoD mechanism directly into a 3DGS representation. Our method introduces a learnable, distance-dependent decay parameter for each Gaussian primitive, which dynamically adjusts its opacity based on viewpoint proximity. This allows for the progressive and smooth filtering of less significant primitives, effectively creating a continuous spectrum of detail within one model. To train this model to be robust across all distances, we introduce a virtual distance scaling mechanism and a novel coarse-to-fine training strategy with rendered point count regularization. Our approach not only eliminates the storage overhead and visual artifacts of discrete methods but also reduces the primitive count and memory footprint of the final model. Extensive experiments demonstrate that CLoD-GS achieves smooth, quality-scalable rendering from a single model, delivering high-fidelity results across a wide range of performance targets.

Problem

• Eliminating storage overhead of discrete Level-of-Detail models
• Removing visual popping artifacts during detail transitions
• Enabling continuous quality scaling within a single 3D model

Innovation

• Continuous LoD via 3D Gaussian Splatting representation
• Learnable distance-dependent decay parameter for Gaussians
• Virtual distance scaling with coarse-to-fine training strategy