Frequency-Aware Gaussian Splatting Decomposition

2025-03-27

AI Summary

3D Gaussian Splatting (3D-GS) achieves high rendering efficiency but lacks frequency interpretability, hindering explicit disentanglement of geometric structure and textural detail. To address this, we propose a frequency-aware 3D Gaussian lattice decomposition framework: leveraging a Laplacian pyramid, Gaussian ellipsoids are grouped by frequency bands—enabling the first explicit separation of low-frequency geometry from high-frequency appearance. Our method introduces subband-aware grouping, dual-domain (geometry + color) collaborative regularization, and ±color residual modeling. It supports progressive coarse-to-fine optimization and hierarchical rendering. This yields the first editable, hierarchical, and frequency-decoupled explicit 3D-GS representation. Experiments demonstrate significant improvements in editing accuracy and rendering efficiency for band-level style transfer, dynamic LOD control, foveated rendering, and real-time geometric interaction—outperforming state-of-the-art baselines in both controllability and flexibility.

Technology Category

• Computer Vision: 3D Computer Vision
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

3D Gaussian Splatting (3D-GS) has revolutionized novel view synthesis with its efficient, explicit representation. However, it lacks frequency interpretability, making it difficult to separate low-frequency structures from fine details. We introduce a frequency-decomposed 3D-GS framework that groups 3D Gaussians that correspond to subbands in the Laplacian Pyrmaids of the input images. Our approach enforces coherence within each subband (i.e., group of 3D Gaussians) through dedicated regularization, ensuring well-separated frequency components. We extend color values to both positive and negative ranges, allowing higher-frequency layers to add or subtract residual details. To stabilize optimization, we employ a progressive training scheme that refines details in a coarse-to-fine manner. Beyond interpretability, this frequency-aware design unlocks a range of practical benefits. Explicit frequency separation enables advanced 3D editing and stylization, allowing precise manipulation of specific frequency bands. It also supports dynamic level-of-detail control for progressive rendering, streaming, foveated rendering and fast geometry interaction. Through extensive experiments, we demonstrate that our method provides improved control and flexibility for emerging applications in scene editing and interactive rendering. Our code will be made publicly available.

Problem

• Lack of frequency interpretability in 3D Gaussian Splatting representation
• Difficulty separating low-frequency structures from fine details
• Need for advanced 3D editing and dynamic level-of-detail control

Innovation

• Frequency-decomposed 3D-GS framework for subband grouping
• Extended color values for residual detail manipulation
• Progressive training scheme for stable optimization