Authors:Yuke Li, Weihang Liu, Cheng Zhang, Yuefeng Zhang, Jiadi Cui, Zixuan Wang, Junran Ding, Haoyu Wu, Yujiao Shi, Jingyi Yu, Xin Lou
Abstract:Feed-forward 3D Gaussian Splatting models offer fast single-pass reconstruction,but scaling them to match per-scene optimization quality is fundamentally hindered by the scarcity of large-scale 3D annotations. A practical compromise is predict-then-refine,where post-prediction optimization compensates for the limited capacity of the feed-forward network. However,standard feed-forward 3DGS is trained solely for zero-step rendering error,ignoring whether its output constitutes a good initialization for the downstream optimizer. We present ForeSplat,an optimization-aware training framework that equips feed-forward 3DGS models to produce initializations explicitly designed for rapid,effective refinement. By offloading part of the scene-modeling burden to the optimizer,ForeSplat substantially reduces the capacity pressure on the feed-forward model,making high-quality reconstruction feasible even with compact networks. At its core is MetaGrad,a lightweight multi-anchor meta-gradient training rule that bypasses costly higher-order differentiation through the 3DGS optimizer. MetaGrad unrolls a short inner-loop refinement trajectory,samples anchor states,and back-propagates aggregated first-order gradients to the prediction head as a surrogate optimization-aware signal. This fine-tuning adds no inference cost and enables high-quality reconstruction within seconds after a few refinement steps. We instantiate ForeSplat on diverse backbones,including AnySplat,Pi3X,and a distilled variant tailored for edge deployment. Across all tested architectures,a ForeSplat-trained initialization converges in fewer refinement steps and reaches a higher peak reconstruction quality than its vanilla counterpart,even fully converged. The framework consistently bridges the gap between amortized prediction and per-scene optimization,establishing a practical path toward lightweight,high-fidelity 3D reconstruction.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2605.22020 [cs.CV] |
| (or arXiv:2605.22020v2 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2605.22020 arXiv-issued DOI via DataCite |
From: Yuke Li [view email]
[v1]
Thu, 21 May 2026 05:38:47 UTC (12,267 KB)
[v2]
Fri, 22 May 2026 02:28:30 UTC (12,267 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。