GauStudio: A Modular Framework for 3D Gaussian Splatting and Beyond
This work provides tools for researchers and practitioners in 3D computer vision to more easily customize and enhance 3DGS pipelines, though it appears incremental as it builds on existing 3DGS methods.
The authors tackled the problem of improving 3D Gaussian Splatting (3DGS) for novel view synthesis and surface reconstruction by developing GauStudio, a modular framework with a hybrid Gaussian representation and a render-then-fuse approach called GauS, which reduces artifacts in unbounded outdoor scenes and enables high-fidelity mesh reconstruction without fine-tuning.
We present GauStudio, a novel modular framework for modeling 3D Gaussian Splatting (3DGS) to provide standardized, plug-and-play components for users to easily customize and implement a 3DGS pipeline. Supported by our framework, we propose a hybrid Gaussian representation with foreground and skyball background models. Experiments demonstrate this representation reduces artifacts in unbounded outdoor scenes and improves novel view synthesis. Finally, we propose Gaussian Splatting Surface Reconstruction (GauS), a novel render-then-fuse approach for high-fidelity mesh reconstruction from 3DGS inputs without fine-tuning. Overall, our GauStudio framework, hybrid representation, and GauS approach enhance 3DGS modeling and rendering capabilities, enabling higher-quality novel view synthesis and surface reconstruction.