Optimized Minimal 4D Gaussian Splatting

OMG4
Optimzied Minimal 4D Gaussian Splatting

Minseo Lee¹^*, Byeonghyeon Lee¹^*, Lucas Yunkyu Lee^2,3, Eunsoo Lee⁴, Sangmin Kim², Seunghyeon Song⁴, Joo Chan Lee⁴, Jong Hwan Ko⁴, Jaesik Park², and Eunbyung Park¹,

¹Yonsei University ²Seoul National University ³POSTECH ⁴Sungkyunkwan University
^*Equal contribution

Abstract

4D Gaussian Splatting has emerged as a new paradigm for dynamic scene representation, enabling real-time rendering of scenes with complex motions. However, it faces a major challenge of storage overhead, as millions of Gaussians are required for high-fidelity reconstruction. While several studies have attempted to alleviate this memory burden, they still face limitations in compression ratio or visual quality. In this work, we present OMG4 (Optimized Minimal 4D Gaussian Splatting), a framework that constructs a compact set of salient Gaussians capable of faithfully representing 4D Gaussian models. Our method progressively prunes Gaussians in three stages: (1) Gaussian Sampling to identify primitives critical to reconstruction fidelity, (2) Gaussian Pruning to remove redundancies, and (3) Gaussian Merging to fuse primitives with similar characteristics. In addition, we integrate implicit appearance compression and generalize Sub-Vector Quantization (SVQ) to 4D representations, further reducing storage while preserving quality. Extensive experiments on standard benchmark datasets demonstrate that OMG4 significantly outperforms recent state-of-the-art methods, reducing model sizes by over 60% while maintaining reconstruction quality. These results position OMG4 as a significant step forward in compact 4D scene representation, opening new possibilities for a wide range of applications.

Optimized Minimal 4D Gaussians

We progressively prune 4D Gaussian primitives to minimize redundancy while preserving fidelity. To achieve a efficiently compressed yet high quality model, OMG4 identifies the most crucial primitives, removes redundancies, and fuses those with similar characteristics. Combining this process with attribute compression, we ensure efficient 4D scene reconstruction.

Gaussian Sampling

To sample the most crucial Gaussians, we calculate their importance using the SD score. Gaussian Sampling selects those that have the greatest impact on rendering quality, preserving PSNR while retaining only 20% of the original representation.

Gaussian Pruning

Although the initial Gaussian Sampling stage with the SD-Score produces a set of critical Gaussians, it still contains redundancies. To address this, Gaussian Pruning strategy further refines the set by removing unnecessary Gaussians.

Visual Comparisons