Variation-aware Flexible 3D Gaussian Editing

2026-02-12

AI Summary

Existing indirect editing methods for 3D Gaussian splatting rely on back-projecting 2D manipulations into 3D space, often leading to multi-view inconsistencies that limit editing flexibility and efficiency. This work proposes VF-Editor, the first end-to-end framework for native 3D Gaussian editing. VF-Editor introduces a learnable parallel decoding function to construct a change predictor that distills unified knowledge from diverse 2D editors, generating a change field to iteratively infer adjustments for individual Gaussian attributes. By enabling knowledge fusion from various 2D editing strategies, the method significantly enhances consistency, flexibility, and efficiency in 3D editing. Experiments demonstrate that VF-Editor effectively mitigates view inconsistency on both public and private datasets, achieving high-quality 3D edits.

Technology Category

• Computer Vision: 3D Computer Vision
• Machine Learning: Learning with Manifolds
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Indirect editing methods for 3D Gaussian Splatting (3DGS) have recently witnessed significant advancements. These approaches operate by first applying edits in the rendered 2D space and subsequently projecting the modifications back into 3D. However, this paradigm inevitably introduces cross-view inconsistencies and constrains both the flexibility and efficiency of the editing process. To address these challenges, we present VF-Editor, which enables native editing of Gaussian primitives by predicting attribute variations in a feedforward manner. To accurately and efficiently estimate these variations, we design a novel variation predictor distilled from 2D editing knowledge. The predictor encodes the input to generate a variation field and employs two learnable, parallel decoding functions to iteratively infer attribute changes for each 3D Gaussian. Thanks to its unified design, VF-Editor can seamlessly distill editing knowledge from diverse 2D editors and strategies into a single predictor, allowing for flexible and effective knowledge transfer into the 3D domain. Extensive experiments on both public and private datasets reveal the inherent limitations of indirect editing pipelines and validate the effectiveness and flexibility of our approach.

Problem

• 3D Gaussian Splatting
• indirect editing
• cross-view inconsistency
• editing flexibility
• editing efficiency

Innovation

• 3D Gaussian Splatting
• direct editing
• variation predictor
• knowledge distillation
• feedforward editing