Geometry Gaussians: Decoupling Appearance and Geometry in Gaussian Splatting
For researchers in novel view synthesis and 3D reconstruction, this work provides a simple fix to a known limitation of 3DGS, enabling better geometry extraction without sacrificing appearance quality.
The paper identifies that 3D Gaussian Splatting (3DGS) inherently struggles to simultaneously represent texture and geometry, and proposes adding a geometry opacity parameter to each splat along with a transparency-curated optimization pipeline. This yields improved rendering and geometry performance across diverse datasets, particularly for scenes with transparent objects.
After the success of 3D Gaussian Splatting (3DGS) for novel view synthesis, many works have explored how to also use it for geometric surface representation. However, extracting accurate geometric information directly from 3DGS remains challenging and can often reduce the appearance rendering quality. In this work, we show that 3DGS in its default form is inheritedly unsuited to represent texture and geometry at the same time, by training with complete ground-truth texture and geometry information. We also propose a simple solution by applying a single additional geometry opacity parameter to each splat, together with an optional transparency-curated optimization pipeline. Our experiments, both with ground-truth and vision foundation model geometric input, show that this change leads to improved rendering and geometry performance on a wide variety of dataset, and especially complex scenes with transparent objects benefit significantly from our method.