4D Scaffold Gaussian Splatting with Dynamic-Aware Anchor Growing for Efficient and High-Fidelity Dynamic Scene Reconstruction

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Cho, Woong Oh, Cho, In, Kim, Seoha, Bae, Jeongmin, Uh, Youngjung, Kim, Seon Joo
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918114886680576
author Cho, Woong Oh
Cho, In
Kim, Seoha
Bae, Jeongmin
Uh, Youngjung
Kim, Seon Joo
author_facet Cho, Woong Oh
Cho, In
Kim, Seoha
Bae, Jeongmin
Uh, Youngjung
Kim, Seon Joo
contents Modeling dynamic scenes through 4D Gaussians offers high visual fidelity and fast rendering speeds, but comes with significant storage overhead. Recent approaches mitigate this cost by aggressively reducing the number of Gaussians. However, this inevitably removes Gaussians essential for high-quality rendering, leading to severe degradation in dynamic regions. In this paper, we introduce a novel 4D anchor-based framework that tackles the storage cost in different perspective. Rather than reducing the number of Gaussians, our method retains a sufficient quantity to accurately model dynamic contents, while compressing them into compact, grid-aligned 4D anchor features. Each anchor is processed by an MLP to spawn a set of neural 4D Gaussians, which represent a local spatiotemporal region. We design these neural 4D Gaussians to capture temporal changes with minimal parameters, making them well-suited for the MLP-based spawning. Moreover, we introduce a dynamic-aware anchor growing strategy to effectively assign additional anchors to under-reconstructed dynamic regions. Our method adjusts the accumulated gradients with Gaussians' temporal coverage, significantly improving reconstruction quality in dynamic regions. Experimental results highlight that our method achieves state-of-the-art visual quality in dynamic regions, outperforming all baselines by a large margin with practical storage costs.
format Preprint
id arxiv_https___arxiv_org_abs_2411_17044
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle 4D Scaffold Gaussian Splatting with Dynamic-Aware Anchor Growing for Efficient and High-Fidelity Dynamic Scene Reconstruction
Cho, Woong Oh
Cho, In
Kim, Seoha
Bae, Jeongmin
Uh, Youngjung
Kim, Seon Joo
Computer Vision and Pattern Recognition
Graphics
Modeling dynamic scenes through 4D Gaussians offers high visual fidelity and fast rendering speeds, but comes with significant storage overhead. Recent approaches mitigate this cost by aggressively reducing the number of Gaussians. However, this inevitably removes Gaussians essential for high-quality rendering, leading to severe degradation in dynamic regions. In this paper, we introduce a novel 4D anchor-based framework that tackles the storage cost in different perspective. Rather than reducing the number of Gaussians, our method retains a sufficient quantity to accurately model dynamic contents, while compressing them into compact, grid-aligned 4D anchor features. Each anchor is processed by an MLP to spawn a set of neural 4D Gaussians, which represent a local spatiotemporal region. We design these neural 4D Gaussians to capture temporal changes with minimal parameters, making them well-suited for the MLP-based spawning. Moreover, we introduce a dynamic-aware anchor growing strategy to effectively assign additional anchors to under-reconstructed dynamic regions. Our method adjusts the accumulated gradients with Gaussians' temporal coverage, significantly improving reconstruction quality in dynamic regions. Experimental results highlight that our method achieves state-of-the-art visual quality in dynamic regions, outperforming all baselines by a large margin with practical storage costs.
title 4D Scaffold Gaussian Splatting with Dynamic-Aware Anchor Growing for Efficient and High-Fidelity Dynamic Scene Reconstruction
topic Computer Vision and Pattern Recognition
Graphics
url https://arxiv.org/abs/2411.17044