Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

NanoGS: Training-Free Gaussian Splat Simplification

Butian Xiong, Rong Liu, Tiantian Zhou, Meida Chen, Zhiwen Fan, Andrew Feng

Abstract

3D Gaussian Splat (3DGS) enables high-fidelity, real-time novel view synthesis by representing scenes with large sets of anisotropic primitives, but often requires millions of Splats, incurring significant storage and transmission costs. Most existing compression methods rely on GPU-intensive post-training optimization with calibrated images, limiting practical deployment. We introduce NanoGS, a training-free and lightweight framework for Gaussian Splat simplification. Instead of relying on image-based rendering supervision, NanoGS formulates simplification as local pairwise merging over a sparse spatial graph. The method approximates a pair of Gaussians with a single primitive using mass preserved moment matching and evaluates merge quality through a principled merge cost between the original mixture and its approximation. By restricting merge candidates to local neighborhoods and selecting compatible pairs efficiently, NanoGS produces compact Gaussian representations while preserving scene structure and appearance. NanoGS operates directly on existing Gaussian Splat models, runs efficiently on CPU, and preserves the standard 3DGS parameterization, enabling seamless integration with existing rendering pipelines. Experiments demonstrate that NanoGS substantially reduces primitive count while maintaining high rendering fidelity, providing an efficient and practical solution for Gaussian Splat simplification. Our project website is available at https://saliteta.github.io/NanoGS/.

NanoGS: Training-Free Gaussian Splat Simplification

Abstract

3D Gaussian Splat (3DGS) enables high-fidelity, real-time novel view synthesis by representing scenes with large sets of anisotropic primitives, but often requires millions of Splats, incurring significant storage and transmission costs. Most existing compression methods rely on GPU-intensive post-training optimization with calibrated images, limiting practical deployment. We introduce NanoGS, a training-free and lightweight framework for Gaussian Splat simplification. Instead of relying on image-based rendering supervision, NanoGS formulates simplification as local pairwise merging over a sparse spatial graph. The method approximates a pair of Gaussians with a single primitive using mass preserved moment matching and evaluates merge quality through a principled merge cost between the original mixture and its approximation. By restricting merge candidates to local neighborhoods and selecting compatible pairs efficiently, NanoGS produces compact Gaussian representations while preserving scene structure and appearance. NanoGS operates directly on existing Gaussian Splat models, runs efficiently on CPU, and preserves the standard 3DGS parameterization, enabling seamless integration with existing rendering pipelines. Experiments demonstrate that NanoGS substantially reduces primitive count while maintaining high rendering fidelity, providing an efficient and practical solution for Gaussian Splat simplification. Our project website is available at https://saliteta.github.io/NanoGS/.
Paper Structure (18 sections, 17 equations, 6 figures, 6 tables, 1 algorithm)

This paper contains 18 sections, 17 equations, 6 figures, 6 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Works
  3. 3DGS Compression: Bit-Level Storage Reduction
  4. 3DGS Compaction: Primitive-Level Count Reduction
  5. Methods
  6. Sparse Merge Graph Construction
  7. Merge Cost
  8. Merging Operation
  9. Progressive Batched Merging
  10. Experiments
  11. Comparison with the State of the Art
  12. Ablation Study
  13. Conclusion
  14. Special Case for I Divergence
  15. Analytical cases and practical estimator.
  16. ...and 3 more sections

Figures (6)

  • Figure 1: NanoGS reduces Gaussian Splat primitive count from the full model to increasingly compact representations. NanoGS achieves substantial compaction ratio while preserving visual fidelity and geometric structure without GPU-intensive optimization or calibrated images.
  • Figure 2: NanoGS pipeline overview.(Top) Starting from an initial set of Gaussian splats, NanoGS constructs a sparse $k$-nearest-neighbor graph, evaluates a merge cost for each candidate edge, and collapses the lowest-cost disjoint pairs progressively. (Bottom-left) The merge cost combines an appearance term and a geometry term. The geometry cost measures the I-divergence between the original two-splat mixture and its single-Gaussian approximation. (Bottom-right) The Mass Preserved Moment Matching(MPMM) merge operator fuses a selected pair by computing mass-weighted moments, so the merged Gaussian is biased toward the larger primitive. Opacity is aggregated as a probabilistic union, and appearance features are blended as a weighted average.
  • Figure 3: Qualitative comparison on chair from NeRF Synthetic dataset. We show two representative test views. Columns correspond to the compaction ratio $\rho\in\{0.1,0.01,0.001\}$ (leftmost: ground truth), and rows compare LightGS, PUP3DGS, GHAP, and Ours.
  • Figure 4: Qualitative comparison on Mip-NeRF 360 dataset. We show two representative test views. Columns correspond to the compaction ratio $\rho\in\{0.1,0.01,0.001\}$ (leftmost: ground truth), and rows compare LightGS, PUP3DGS, GHAP, and Ours.
  • Figure 5: Generated results and corresponding compression results
  • ...and 1 more figures