Robust 3D-Masked Part-level Editing in 3D Gaussian Splatting with Regularized Score Distillation Sampling
arXiv:2507.11061v3 Announce Type: replace-cross Abstract: Recent advances in 3D neural representations and instance-level editing models have enabled the efficient creation of high-quality 3D content. However, achieving precise local 3D edits remains challenging, especially for Gaussian Splatting, due to inconsistent multi-view 2D part segmentations and inherently ambiguous nature of Score Distillation Sampling (SDS) loss. To address these limitations, we propose RoMaP, a novel local 3D Gaussian editing framework that enables precise and drastic part-level modifications. First, we introduce a
The paper addresses current limitations in 3D content creation, particularly precise local editing in Gaussian Splatting, which is a rapidly evolving area in 3D AI research.
Precise and drastic part-level modifications in 3D Gaussian Splatting can significantly improve the efficiency and quality of 3D content generation, impacting industries from gaming to product design.
The ability to perform robust part-level editing in 3D Gaussian Splatting overcomes previous inconsistencies and ambiguities, allowing for more granular and controllable 3D asset creation.
- · 3D content creators
- · Gaming industry
- · E-commerce platforms
- · AI model developers
- · Manual 3D editing workflows
- · Less precise 3D generation techniques
More sophisticated and customized 3D models can be generated with less effort and higher fidelity.
This improved 3D content generation could accelerate the development of virtual worlds and digital twins.
Enhanced 3D interaction capabilities might lead to new forms of human-computer interaction and product prototyping.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI