Search papers, labs, and topics across Lattice.
The paper introduces GSCompleter, a novel plugin for 3D Gaussian Splatting (3DGS) that addresses performance degradation under sparse-view extrapolation by shifting from an unstable "Repair-then-Distill" paradigm to a stable "Generate-then-Register" workflow. GSCompleter synthesizes 2D reference images, lifts them into metric-scale 3D primitives using a Stereo-Anchor mechanism, and integrates them into the global context via a Ray-Constrained Registration strategy. The method achieves state-of-the-art 3DGS completion performance across multiple benchmarks, improving both quality and efficiency.
Ditch the unstable "Repair-then-Distill" paradigm: GSCompleter's "Generate-then-Register" workflow completes 3D Gaussian Splattings in seconds, setting a new SOTA in sparse-view extrapolation.
While 3D Gaussian Splatting (3DGS) has revolutionized real-time rendering, its performance degrades significantly under sparse-view extrapolation, manifesting as severe geometric voids and artifacts. Existing solutions primarily rely on an iterative"Repair-then-Distill"paradigm, which is inherently unstable and prone to overfitting. In this work, we propose GSCompleter, a distillation-free plugin that shifts scene completion to a stable"Generate-then-Register"workflow. Our approach first synthesizes plausible 2D reference images and explicitly lifts them into metric-scale 3D primitives via a robust Stereo-Anchor mechanism. These primitives are then seamlessly integrated into the global context through a novel Ray-Constrained Registration strategy. This shift to a rapid registration paradigm delivers superior 3DGS completion performance across three distinct benchmarks, enhancing the quality and efficiency of various baselines and achieving new SOTA results.
