TexEditor: Structure-Preserving Text-Driven Texture Editing

2026-03-19

AI Summary

This work addresses the common issue of appearance distortion in text-driven texture editing caused by neglecting underlying geometric structure. To this end, we propose StructureNFT, a method that jointly optimizes data generation and model training by integrating a synthetic dataset, TexBlender, with a reinforcement learning strategy to significantly enhance structural fidelity. We further introduce TexBench, the first benchmark tailored for evaluating texture editing in real-world scenarios. Building upon the Qwen-Image-Edit-2509 model, our approach combines synthetic data, supervised fine-tuning, and reinforcement learning. Experiments demonstrate that StructureNFT outperforms existing methods—including Nano Banana Pro—across multiple synthetic and real-world benchmarks, while also exhibiting strong generalization capabilities on the general-purpose image editing benchmark ImgEdit.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Structured Learning

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Text-guided texture editing aims to modify object appearance while preserving the underlying geometric structure. However, our empirical analysis reveals that even SOTA editing models frequently struggle to maintain structural consistency during texture editing, despite the intended changes being purely appearance-related. Motivated by this observation, we jointly enhance structure preservation from both data and training perspectives, and build TexEditor, a dedicated texture editing model based on Qwen-Image-Edit-2509. Firstly, we construct TexBlender, a high-quality SFT dataset generated with Blender, which provides strong structural priors for a cold start. Sec- ondly, we introduce StructureNFT, a RL-based approach that integrates structure-preserving losses to transfer the structural priors learned during SFT to real-world scenes. Moreover, due to the limited realism and evaluation coverage of existing benchmarks, we introduce TexBench, a general-purpose real-world benchmark for text-guided texture editing. Extensive experiments on existing Blender-based texture benchmarks and our TexBench show that TexEditor consistently outperforms strong baselines such as Nano Banana Pro. In addition, we assess TexEditor on the general purpose benchmark ImgEdit to validate its generalization. Our code and data are available at https://github.com/KlingAIResearch/TexEditor.

Problem

• text-guided texture editing
• structure preservation
• geometric consistency
• texture editing

Innovation

• structure-preserving
• text-driven texture editing
• SFT dataset
• reinforcement learning
• real-world benchmark