Adaptive texture allocation cuts storage for 4d gaussian splatting
ADATEX4D: adaptive texture capacity allocation for 4D gaussian splatting
Computer Vision and Pattern RecognitionArtificial Intelligence
Summary
When creating 3D scenes that change over time, some parts of the scene need detailed textures while others do not. The authors introduce a way to assign texture details unevenly, giving more detail where it matters and less where it doesn't. This method, AdaTex4D, saves a lot of memory without losing image quality. It works by looking at how visible and detailed different parts of the scene are and adjusting texture sizes accordingly.
What this means in practice
- •For 3d graphics developers: Build 4D Gaussian Splatting models that use less texture memory without sacrificing visual quality by adapting texture resolution to scene detail and visibility.
- •For augmented reality engineers: Create more efficient dynamic 3D scene representations that reduce memory footprint, enabling smoother real-time rendering on limited hardware.
Authors
De Jiang, Peiqiang Wang, Kehong Yuan, Shaohua Ma
Abstract
Textured Gaussians improve local appearance capacity, but assigning the same texture resolution to every primitive wastes storage on low-detail or weakly visible regions. We introduce AdaTex4D, an adaptive texture-capacity module for deformation-based 4D Gaussian Splatting. Each Gaussian carries packed RGBA triplanes whose two axes grow independently according to visibility normalized screen-space gradients and deformed local scales. Experiments on N3DV and PanopticSports show that AdaTex4D reduces texture storage by more than half while preserving reconstruction quality. Under fixed memory budgets, adaptive allocation also improves quality over uniform texture assignment and reduces overall model and peak memory. These results show that dynamic, anisotropic texture allocation provides a more efficient way to distribute local appearance capacity in 4D Gaussian representations.