Problem
Research questions and friction points this paper is trying to address.
3D material generation
texture detail fidelity
texture reconstruction
high-frequency textures
TaoTex: Boosting Texture Detail Fidelity for Native 3D Material Generation
🤖 AI Summary
This study addresses the persistent challenge of reconstructing fine-grained textures in existing 3D generative models. We propose a diffusion-based framework for native 3D material generation that enhances training samples by constructing high-frequency texture data proxies. The architecture incorporates a multi-level feature fusion module coupled with a latent-to-pixel loss transition strategy, and introduces learnable view embeddings to ensure multi-view consistency. Extensive evaluations demonstrate that the proposed method significantly outperforms existing baselines under both single- and multi-view conditions, substantially improving the fidelity of reconstructed 3D texture details and overall generation quality.
📝 Abstract
Recent 3D generation models can produce accurate geometries while still struggling to reconstruct detailed textures. We propose a diffusion-based native 3D material generation model TaoTex, which faithfully recovers intricate textures through tailored strategies and improvements. First, we develop a data construction agent to create high-frequency textured 3D assets to bridge the data gap in public datasets. Training with these data significantly enhances the ability of TaoTex to recover challenging details such as text and patterns. Second, we design a multi-level feature fusion (MLFF) module to adaptively integrate local and global features of the conditional input, providing more complete texture cues for the diffusion model and thereby enhancing reconstruction fidelity. To alleviate VAE reconstruction errors, we adopt a latent-to-pixel space loss transition, further improving the pixel-level details and generation quality. Finally, we scale TaoTex to multi-view inputs by incorporating learnable viewpoint embeddings, achieving accurate and consistent material reconstruction across views. Extensive experiments demonstrate that our method significantly outperforms existing approaches in preserving texture details in both single- and multi-view settings.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Native 3D Material Generation
Multi-level Feature Fusion
Latent-to-Pixel Space Loss
Data Construction Agent
Learnable Viewpoint Embeddings
🔎 Similar Papers
No similar papers found.
