Prox-E: Fine-Grained 3D Shape Editing via Primitive-Based Abstractions

2026-04-26

AI Summary

Existing 2D image-driven 3D editing methods struggle to achieve fine-grained local modifications while preserving the overall identity of the object. This work proposes a training-free 3D editing framework that introduces, for the first time, an explicit geometry-primitive-level abstraction mechanism: it decomposes the input 3D shape into a set of geometric primitives, leverages a pre-trained vision-language model to interpret editing instructions and perform semantic operations on these primitives, and subsequently guides a 3D generative model to realize high-fidelity local edits. The proposed method significantly outperforms existing 2D-driven and training-based 3D editing approaches in terms of identity preservation, shape quality, and instruction adherence.

Technology Category

• Computer Vision: 3D Computer Vision
• Natural Language Processing: Language Grounding & Multi-modal NLP
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Text-based 2D image editing models have recently reached an impressive level of maturity, motivating a growing body of work that heavily depends on these models to drive 3D edits. While effective for appearance-based modifications, such 2D-centric 3D editing pipelines often struggle with fine-grained 3D editing, where localized structural changes must be applied while strictly preserving an object's overall identity. To address this limitation, we propose Prox-E, a training-free framework that enables fine-grained 3D control through an explicit, primitive-based geometric abstraction. Our framework first abstracts an input 3D shape into a compact set of geometric primitives. A pretrained vision-language model (VLM) then edits this abstraction to specify primitive-level changes. These structural edits are subsequently used to guide a 3D generative model, enabling fine-grained, localized modifications while preserving unchanged regions of the original shape. Through extensive experiments, we demonstrate that our method consistently balances identity preservation, shape quality, and instruction fidelity more effectively than various existing approaches, including 2D-based 3D editors and training-based methods.

Problem

• fine-grained 3D editing
• 3D shape editing
• identity preservation
• localized structural changes
• primitive-based abstraction

Innovation

• primitive-based abstraction
• fine-grained 3D editing
• training-free framework
• vision-language model
• 3D generative model