Learnable Multi-level Discrete Wavelet Transforms for 3D Gaussian Splatting Frequency Modulation

TL;DR

This paper introduces a multi-level Discrete Wavelet Transform framework for 3D Gaussian Splatting that reduces memory costs by controlling Gaussian primitive growth through adaptive frequency modulation.

Contribution

It extends previous wavelet-based modulation by employing multi-level DWT and a single scaling parameter, improving Gaussian count reduction during training.

Findings

Reduces Gaussian primitive counts compared to prior methods.

Maintains competitive rendering quality with fewer primitives.

Uses a deeper, recursive frequency modulation curriculum.

Abstract

3D Gaussian Splatting (3DGS) has emerged as a powerful approach for novel view synthesis. However, the number of Gaussian primitives often grows substantially during training as finer scene details are reconstructed, leading to increased memory and storage costs. Recent coarse-to-fine strategies regulate Gaussian growth by modulating the frequency content of the ground-truth images. In particular, AutoOpti3DGS employs the learnable Discrete Wavelet Transform (DWT) to enable data-adaptive frequency modulation. Nevertheless, its modulation depth is limited by the 1-level DWT, and jointly optimizing wavelet regularization with 3D reconstruction introduces gradient competition that promotes excessive Gaussian densification. In this paper, we propose a multi-level DWT-based frequency modulation framework for 3DGS. By recursively decomposing the low-frequency subband, we construct a deeper curriculum that provides progressively coarser supervision during early training, consistently reducing Gaussian counts. Furthermore, we show that the modulation can be performed using only a single scaling parameter, rather than learning the full 2-tap high-pass filter. Experimental results on standard benchmarks demonstrate that our method further reduces Gaussian counts while maintaining competitive rendering quality.