We propose a native material generation model with the capability to recover texture details from both single- and multi-view inputs. For training this model, we develop a 3D asset data construction agent based on the Qwen LLM, which enables the automatic creation of diverse 3D assets with detailed text, logo patterns, and various material properties. To better capture fine details from the conditional images, we propose a Multi-level Feature Fusion module that adaptively integrates both semantic and local information, allowing a more faithful reconstruction. To mitigate the accuracy degradation caused by the 3D VAE, we use a two-stage training strategy, starting with latent-space loss and then switching to pixel-space loss. Furthermore, we scale TaoTex to multi-view inputs by introducing learnable viewpoint embeddings. Combined with our constructed dataset and trained with a low-SNR schedule, this enables accurate and consistent texture reconstruction across different views.
Loading Cloth…
Drag to rotate · Right-drag to pan
@misc{wu2026taotex,
title={TaoTex: Boosting Texture Detail Fidelity for Native 3D Material Generation},
author={Xiuchao Wu and Shuichang Lai and Jiangjing Lyu and Chengfei Lyu},
year={2026},
eprint={2609.34934},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2609.34934},
}