GOATex: Geometry & Occlusion-Aware Texturing

TL;DR

GOATex tackles the occlusion-enabled texturing of both exterior and interior mesh surfaces by introducing a geometry-guided, occlusion-aware pipeline that partitions surfaces into ordered hit-level layers. It employs a two-stage visibility control to preserve global shape while progressively revealing interior geometry, textures each layer with depth-conditioned diffusion models, and softly blends textures in UV space using view-dependent weights. The method achieves seamless, high-fidelity textures across visible and occluded regions without finetuning diffusion models and supports separate prompting for inner and outer surfaces for fine-grained stylistic control. Extensive experiments, including human and GPT-based evaluations, demonstrate superior interior texture quality and competitive exterior fidelity relative to baselines, with ablations validating the contribution of each component. This work opens new avenues for controllable, high-quality mesh texturing in AR/VR and game pipelines, while noting limitations in semantic coherence and potential societal risks of diffusion-based texture synthesis.

Abstract

We present GOATex, a diffusion-based method for 3D mesh texturing that generates high-quality textures for both exterior and interior surfaces. While existing methods perform well on visible regions, they inherently lack mechanisms to handle occluded interiors, resulting in incomplete textures and visible seams. To address this, we introduce an occlusion-aware texturing framework based on the concept of hit levels, which quantify the relative depth of mesh faces via multi-view ray casting. This allows us to partition mesh faces into ordered visibility layers, from outermost to innermost. We then apply a two-stage visibility control strategy that progressively reveals interior regions with structural coherence, followed by texturing each layer using a pretrained diffusion model. To seamlessly merge textures obtained across layers, we propose a soft UV-space blending technique that weighs each texture's contribution based on view-dependent visibility confidence. Empirical results demonstrate that GOATex consistently outperforms existing methods, producing seamless, high-fidelity textures across both visible and occluded surfaces. Unlike prior works, GOATex operates entirely without costly fine-tuning of a pretrained diffusion model and allows separate prompting for exterior and interior mesh regions, enabling fine-grained control over layered appearances. For more qualitative results, please visit our project page: https://goatex3d.github.io/.

Figures (17)

• Figure 1: Overall Pipeline of GOATex. (1) Our framework begins with superface construction, grouping fine-grained faces into coherent regions. (2) We then assign each superface a hit level by casting rays from multiple viewpoints, indicating its relative depth within the mesh (§\ref{['sec:preprocessing']}). (3) Based on these hit levels, we apply a visibility control strategy, combining residual face clustering with normal flipping and backface culling, to progressively reveal interior geometry without disrupting the object's structural integrity (§\ref{['sec:3.2']}). (4) For each hit level, we synthesize textures using a depth-conditioned multi-view diffusion (MVD) module liu2023syncmvd. (5) We employ a soft, visibility-weighted UV-blending strategy to merge textures across levels, ensuring seamless and coherent appearance (§\ref{['sec:3.3']}).
• Figure 2: Analysis on our two visibility control techniques, residual face clustering and normal flipping. (Left) Removing both leads to sparse, fragmented face clusters and causes visible seams (e.g., around chair edges). (Middle) Residual clustering improves cluster density but still suffers from fragmentation at hit level 4, suffering from incomplete textures (e.g., floor regions). (Right) In contrast, combining both techniques results in improved visibility segmentation, effectively revealing interior structures while preserving the overall shape and structural coherence of the object.
• Figure 3: Analysis on UV texture merging techniques. Compared to naive approaches, our blending strategy preserves clear structural boundaries and significantly improves texture quality.
• Figure 4: Qualitative comparisons. Our GOATex significantly outperforms all the baselines in interior surface texturing, while maintaining competitive in texturing exterior regions (boxed area at the top right of each object). This balanced approach makes GOATex a robust solution for both visible exterior and occluded interior surface texturing.
• Figure 5: Preference rates of our method over the baselines, as judged by human raters or GPTs.