无需训练即可精准局部3D风格化，解决风格泄漏和边界模糊问题。
Owner3D: Ownership-Guided Style Writing for Training-Free Localized 3D Stylization
- 通过所有权引导写入风格，仅在目标区域注入参考风格。
- 减少86.4%~89.9%的风格泄露，显著提升目标区域保真度。
- 适合需要快速、无训练局部风格化的真实场景应用。
局部3D风格化旨在修改指定物体部分的外观，同时保留其余表面。在大型重建模型（LRMs）中，该任务具有挑战性，因为风格在渲染前被注入中间外观表示，而紧凑的三平面特征在目标与非目标表面间共享，导致风格泄漏和边界模糊。本文提出Owner3D，一种无需训练的局部3D风格化框架，将局部外观控制直接整合进LRM重建过程。具体地，Owner3D引入所有权引导风格写入机制，限制参考风格仅注入目标区域，生成单一局部风格化三平面，且无需额外训练，避免了全局风格与外观表示的分离。为解决语义边界附近的外观模糊问题，进一步引入边界双槽机制，分别维护目标与非目标区域的局部特征源。最后，采用表面优先的纹理读取策略，分层融合表面、3D及三平面所有权证据，以应对不完整可见性下的外观恢复。在基于Google Scanned Objects与PartNet构建的基准上，Owner3D在目标区域风格保真度与非目标外观保持方面持续优于现有方法，相比StyleSplat和LAENeRF，分别降低86.4%与89.9%的外观泄漏。
原文摘要 · Abstract (English)
Localized 3D stylization aims to modify the appearance of a specified object part while preserving the remaining surfaces. In large reconstruction models (LRMs), this task is challenging because style is injected into intermediate appearance representations before rendering, while compact triplane features are shared across target and non-target surfaces, causing style leakage and boundary ambiguity. We propose Owner3D, a training-free framework for localized 3D stylization that integrates localized appearance control directly into the LRM reconstruction process. Specifically, Owner3D introduces ownership-guided style writing to restrict reference-style injection to target regions, producing a single localized stylized triplane without additional training while avoiding separate global style and appearance representations. To resolve appearance ambiguity near semantic boundaries, we further introduce boundary dual slots that maintain separate local feature sources for target and non-target regions. Finally, a surface-first texture readout hierarchically combines surface, 3D, and triplane ownership evidence to robustly recover appearance under incomplete visibility. Experiments on a benchmark constructed from Google Scanned Objects and PartNet demonstrate that Owner3D consistently outperforms existing 3D stylization methods in target-region style fidelity and non-target appearance preservation, reducing appearance leakage by 86.4% and 89.9% compared with StyleSplat and LAENeRF, respectively.
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。