Fully Explicit Dynamic Gaussian Splatting

TL;DR

The key idea is to firstly separate static and dynamic Gaussians during training, and to explicitly sample positions and rotations of the dynamic Gaussians at sparse timestamps to represent both spatially and temporally continuous motions of objects in dynamic scenes as well as reducing computational cost.

Abstract

3D Gaussian Splatting has shown fast and high-quality rendering results in static scenes by leveraging dense 3D prior and explicit representations. Unfortunately, the benefits of the prior and representation do not involve novel view synthesis for dynamic motions. Ironically, this is because the main barrier is the reliance on them, which requires increasing training and rendering times to account for dynamic motions. In this paper, we design a Explicit 4D Gaussian Splatting(Ex4DGS). Our key idea is to firstly separate static and dynamic Gaussians during training, and to explicitly sample positions and rotations of the dynamic Gaussians at sparse timestamps. The sampled positions and rotations are then interpolated to represent both spatially and temporally continuous motions of objects in dynamic scenes as well as reducing computational cost. Additionally, we introduce a progressive training scheme and a point-backtracking technique that improves Ex4DGS's convergence. We initially train Ex4DGS using short timestamps and progressively extend timestamps, which makes it work well with a few point clouds. The point-backtracking is used to quantify the cumulative error of each Gaussian over time, enabling the detection and removal of erroneous Gaussians in dynamic scenes. Comprehensive experiments on various scenes demonstrate the state-of-the-art rendering quality from our method, achieving fast rendering of 62 fps on a single 2080Ti GPU.

Figures (11)

• Figure 1: Overview of our method. We first initialize 3D Gaussians as static, modeling their motion linearly. During optimization, dynamic and static objects are separated based on the amount of predicted motion, and the 3D Gaussians between the selected keyframes are interpolated and rendered.
• Figure 2: Effectiveness of our keyframe interpolation.
• Figure 3: Comparison between the single Gaussian, Gaussian mixture, and our model for temporal opacity modeling.
• Figure 4: Progressive learning of dynamic Gaussians.
• Figure 5: Comparison of our Ex4DGS with other the state-of-the-art dynamic Gaussian splatting methods on Neural 3D Video li2022neural dataset.