SplatFields: Neural Gaussian Splats for Sparse 3D and 4D Reconstruction
This work solves the practical bottleneck of requiring many input views for high-quality scene reconstruction, especially in dynamic scenes where camera arrays are costly, though it appears incremental as it builds on 3DGS.
The paper tackled the challenge of 3D and 4D reconstruction from sparse multi-view images by addressing the suboptimal performance of 3D Gaussian Splatting in such settings, resulting in consistent enhancement of reconstruction quality across static and dynamic scenarios.
Digitizing 3D static scenes and 4D dynamic events from multi-view images has long been a challenge in computer vision and graphics. Recently, 3D Gaussian Splatting (3DGS) has emerged as a practical and scalable reconstruction method, gaining popularity due to its impressive reconstruction quality, real-time rendering capabilities, and compatibility with widely used visualization tools. However, the method requires a substantial number of input views to achieve high-quality scene reconstruction, introducing a significant practical bottleneck. This challenge is especially severe in capturing dynamic scenes, where deploying an extensive camera array can be prohibitively costly. In this work, we identify the lack of spatial autocorrelation of splat features as one of the factors contributing to the suboptimal performance of the 3DGS technique in sparse reconstruction settings. To address the issue, we propose an optimization strategy that effectively regularizes splat features by modeling them as the outputs of a corresponding implicit neural field. This results in a consistent enhancement of reconstruction quality across various scenarios. Our approach effectively handles static and dynamic cases, as demonstrated by extensive testing across different setups and scene complexities.