提出可精准修复与确定性编辑3D高斯资产的新方法
FocusGS: Spatial Delta Layers for Local Repair and Deterministic Editing of Trained 3D Gaussian Assets
- 用空间增量层统一局部修复与确定性编辑
- 修复使目标区域PSNR提升7.91 dB，编辑平均增益11.05 dB
- 支持轻量级可验证维护，适合模型后期精细化调整
3D高斯点阵（3DGS）正从一次性重建转向可交付、可检查、可维护的视觉资产。现有流程聚焦全局重建、训练时密度控制或开放生成编辑，缺乏对已训练资产的精确局部维护。本文提出FocusGS，将局部修复与确定性编辑统一为复合空间增量。修复为纯加法特例：基项为空，仅添加局部高斯基；确定性编辑采用擦除-插入分解（EIF），结合旧内容擦除与新内容插入。FocusGS缓解空间梯度饥饿问题：在93个评估视图上，修复使目标区域PSNR提升7.91 dB。83次确定性编辑试验中，目标区域均改善，平均编辑后PSNR达21.97 dB，平均增益+11.05 dB；在五个公开编辑案例中，FocusGS-EIF达到33.17 dB目标掩码PSNR与0.994目标增量相关性，而两个文本驱动基线无法完成指定更新。FocusGS提供轻量、可验证的3DGS维护算子。
原文摘要 · Abstract (English)
3D Gaussian Splatting (3DGS) is evolving from one-time reconstruction into deliverable, inspectable, and maintainable visual assets. Existing workflows focus on global reconstruction, training-time density control, or open-ended generative editing, leaving trained assets without precise local maintenance. We propose FocusGS, which unifies local repair and deterministic editing as composite spatial deltas. Repair is the purely additive special case: its base-manipulation term is empty, and it adds only local Gaussian bases; deterministic editing uses erase-insert factorization (EIF) to combine old-carrier erasure with new-content insertion. FocusGS addresses spatial gradient starvation: local repair raises target-region PSNR by 7.91 dB over 93 evaluation views. Across all 83 deterministic editing trials, the target ROI improves, with a trial-averaged mean edited ROI PSNR of 21.97 dB and a mean gain of +11.05 dB; across five public editing cases, FocusGS-EIF reaches 33.17 dB Target-mask PSNR and 0.994 Target-delta Correlation, while both text-driven baselines fail to complete the prescribed updates. FocusGS provides a lightweight, verifiable 3DGS maintenance operator.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。