TSOG: A Format For Temporally And Spatially Ordered Gaussians

2026-07-30

AI Summary

This work addresses the inefficiency and large data volume of 4D Gaussian splatting in representation, storage, and transmission by proposing the TSOG format, which introduces a spatiotemporal ordering index into 4D Gaussian representations for the first time. Building upon the spatially ordered SOG framework, TSOG extends to four-dimensional spacetime through temporal attribute modeling and time-encoded geometric and appearance parameters. It supports both discrete and continuous formulations of 4D Gaussian splatting, offering model-agnosticism and scalability. Experimental results demonstrate that TSOG achieves over 90% reduction in file size compared to PLY sequences and FreeTimeGS baselines, with PSNR variations ranging only from −0.42 to +0.85 dB, thereby drastically lowering storage overhead while incurring negligible fidelity loss.

Abstract

We propose Temporally and Spatially Ordered Gaussians (TSOG), a format for efficient representation of 4D Gaussian Splatting (4DGS) content. TSOG extends the Spatially Ordered Gaussians (SOG) framework to the temporal domain by introducing a timeline attribute and temporal parameterization of geometry and appearance attributes. Similar to SOG, TSOG is a lossy format that assigns each Gaussian a unique index and encodes attribute values as index-aligned image data. TSOG is model-agnostic, extensible, and compatible with both discrete and continuous 4DGS representations. Evaluation using a PLYs sequence and FreeTimeGS as baselines, serving as simplistic and state-of-the-art 4DGS representations respectively, shows file size reductions exceeding 90%, with PSNR differences ranging between -0.42 and +0.85 dB. These results demonstrate substantial file size savings with minimal quality degradation, enabling efficient representation, storage, and delivery of dynamic scenes for next-generation 4D content.

Problem

• 4D Gaussian Splatting
• efficient representation
• file size reduction
• dynamic scenes
• storage and delivery

Innovation

• TSOG
• 4D Gaussian Splatting
• temporal parameterization
• spatial ordering
• efficient compression