MMGS: 10$\times$ Compressed 3DGS through Optimal Transport Aggregation based on Multi-view Ranking

2026-05-18

AI Summary

This work addresses the high computational overhead of 3D Gaussian Splatting caused by redundant primitives and the difficulty of existing compression methods in balancing rendering fidelity and efficiency. The authors formulate Gaussian optimization as a global geometric distribution matching problem and introduce a primitive contribution ranking mechanism grounded in multi-view geometric consistency. By integrating optimal transport–driven global aggregation with a distribution-preserving densification strategy, the approach overcomes the limitations of local heuristic-based compression. Remarkably, the method achieves state-of-the-art rendering quality using only 10% of the original primitives while accelerating training by an order of magnitude.

Abstract

While 3D Gaussian Splatting (3DGS) has revolutionized 3D reconstruction, it suffers from significant overhead due to massive redundant primitives. Existing compression methods typically rely on local sampling or fixed pruning thresholds, which often struggle to balance redundancy reduction with high-fidelity rendering. To address this, we propose a novel framework that formulates Gaussian optimization as a global geometric distribution matching problem. Specifically, our approach integrates three components: (1) we introduce a multi-view 3D Gaussian contribution ranking mechanism that filters primitives using geometric consistency instead of local heuristics; (2) we propose a global Optimal Transport (OT)-based aggregation algorithm that merges redundant primitives while preserving the underlying geometry; and (3) we design an OT-based densification operator that maintains the Gaussian's distributional properties for stable optimization. Our approach achieves state-of-the-art rendering quality with only \textbf{10$\%$} primitives and \textbf{10$\times$} accelerated training speeds compared to vanilla 3DGS.

Problem

• 3D Gaussian Splatting
• redundancy
• compression
• high-fidelity rendering
• primitives

Innovation

• Optimal Transport
• 3D Gaussian Splatting
• Multi-view Ranking
• Geometry-aware Compression
• Global Aggregation