Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2603.19121 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914409786376192 |
|---|---|
| author | Chen, Weilin Rao, Jiahao Wang, Wenhao Li, Xinyang Cheng, Xuan Cao, Liujuan |
| author_facet | Chen, Weilin Rao, Jiahao Wang, Wenhao Li, Xinyang Cheng, Xuan Cao, Liujuan |
| contents | The creation of high-fidelity, customizable 3D indoor scene textures remains a significant challenge. While text-driven methods offer flexibility, they lack the precision for fine-grained, instance-level control, and often produce textures with insufficient quality, artifacts, and baked-in shading. To overcome these limitations, we introduce CustomTex, a novel framework for instance-level, high-fidelity scene texturing driven by reference images. CustomTex takes an untextured 3D scene and a set of reference images specifying the desired appearance for each object instance, and generates a unified, high-resolution texture map. The core of our method is a dual-distillation approach that separates semantic control from pixel-level enhancement. We employ semantic-level distillation, equipped with an instance cross-attention, to ensure semantic plausibility and ``reference-instance'' alignment, and pixel-level distillation to enforce high visual fidelity. Both are unified within a Variational Score Distillation (VSD) optimization framework. Experiments demonstrate that CustomTex achieves precise instance-level consistency with reference images and produces textures with superior sharpness, reduced artifacts, and minimal baked-in shading compared to state-of-the-art methods. Our work establishes a more direct and user-friendly path to high-quality, customizable 3D scene appearance editing. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_19121 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | CustomTex: High-fidelity Indoor Scene Texturing via Multi-Reference Customization Chen, Weilin Rao, Jiahao Wang, Wenhao Li, Xinyang Cheng, Xuan Cao, Liujuan Computer Vision and Pattern Recognition Artificial Intelligence The creation of high-fidelity, customizable 3D indoor scene textures remains a significant challenge. While text-driven methods offer flexibility, they lack the precision for fine-grained, instance-level control, and often produce textures with insufficient quality, artifacts, and baked-in shading. To overcome these limitations, we introduce CustomTex, a novel framework for instance-level, high-fidelity scene texturing driven by reference images. CustomTex takes an untextured 3D scene and a set of reference images specifying the desired appearance for each object instance, and generates a unified, high-resolution texture map. The core of our method is a dual-distillation approach that separates semantic control from pixel-level enhancement. We employ semantic-level distillation, equipped with an instance cross-attention, to ensure semantic plausibility and ``reference-instance'' alignment, and pixel-level distillation to enforce high visual fidelity. Both are unified within a Variational Score Distillation (VSD) optimization framework. Experiments demonstrate that CustomTex achieves precise instance-level consistency with reference images and produces textures with superior sharpness, reduced artifacts, and minimal baked-in shading compared to state-of-the-art methods. Our work establishes a more direct and user-friendly path to high-quality, customizable 3D scene appearance editing. |
| title | CustomTex: High-fidelity Indoor Scene Texturing via Multi-Reference Customization |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2603.19121 |