pith. sign in

arxiv: 2503.12307 · v1 · pith:I5JWE3E2new · submitted 2025-03-16 · 💻 cs.CV · cs.AI

Swift4D:Adaptive divide-and-conquer Gaussian Splatting for compact and efficient reconstruction of dynamic scene

classification 💻 cs.CV cs.AI
keywords dynamicprimitivesstaticdivide-and-conquergaussianmethodonlyquality
0
0 comments X
read the original abstract

Novel view synthesis has long been a practical but challenging task, although the introduction of numerous methods to solve this problem, even combining advanced representations like 3D Gaussian Splatting, they still struggle to recover high-quality results and often consume too much storage memory and training time. In this paper we propose Swift4D, a divide-and-conquer 3D Gaussian Splatting method that can handle static and dynamic primitives separately, achieving a good trade-off between rendering quality and efficiency, motivated by the fact that most of the scene is the static primitive and does not require additional dynamic properties. Concretely, we focus on modeling dynamic transformations only for the dynamic primitives which benefits both efficiency and quality. We first employ a learnable decomposition strategy to separate the primitives, which relies on an additional parameter to classify primitives as static or dynamic. For the dynamic primitives, we employ a compact multi-resolution 4D Hash mapper to transform these primitives from canonical space into deformation space at each timestamp, and then mix the static and dynamic primitives to produce the final output. This divide-and-conquer method facilitates efficient training and reduces storage redundancy. Our method not only achieves state-of-the-art rendering quality while being 20X faster in training than previous SOTA methods with a minimum storage requirement of only 30MB on real-world datasets. Code is available at https://github.com/WuJH2001/swift4d.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Online Segment 3D Gaussians via Launching Virtual Drones

    cs.CV 2026-07 unverdicted novelty 7.0

    SAGO achieves setup-free interactive 3D Gaussian segmentation by modeling it as an online NBV planning task in a Markov process, delivering sub-second latency and over 50x speedup over prior setup-free methods.

  2. PaMoSplat: Part-Aware Motion-Guided Gaussian Splatting for Dynamic Scene Reconstruction

    cs.CV 2026-05 unverdicted novelty 7.0

    PaMoSplat reconstructs dynamic scenes by lifting 2D segmentations to coherent 3D Gaussian parts and estimating their motions via optical flow-guided differential evolution for higher quality rendering and faster training.

  3. DP-DeGauss: Dynamic Probabilistic Gaussian Decomposition for Egocentric 4D Scene Reconstruction

    cs.CV 2026-04 unverdicted novelty 6.0

    DP-DeGauss uses probabilistic Gaussian decomposition with dynamic routing and category masks to achieve the first explicit disentanglement of background, hand, and object components in egocentric 4D scene reconstructi...

  4. LiveStre4m: Feed-Forward Live Streaming of Novel Views from Unposed Multi-View Video

    cs.CV 2026-04 unverdicted novelty 6.0

    LiveStre4m delivers real-time novel-view video streaming from unposed multi-view inputs via a multi-view vision transformer, diffusion-transformer interpolation, and a learned camera pose predictor.

  5. GeoRect4D: Geometry-Compatible Generative Rectification for Dynamic Sparse-View 3D Reconstruction

    cs.CV 2026-04 unverdicted novelty 5.0

    GeoRect4D couples 3D Gaussian splatting with a single-step diffusion rectifier via degradation-aware feedback and progressive optimization to improve fidelity and consistency in sparse-view dynamic 3D reconstruction.