📝 Publications

Kiss3DGen: Repurposing Image Diffusion Models for 3D Asset Generation

📄 Paper (arXiv)
Jiantao Lin*, Xin Yang*, Meixi Chen*, Yingjie Xu, Dongyu Yan, Leyi Wu, Xinli Xu, Lie Xu, Shunsi Zhang, Ying-Cong Chen

Project Page | Code | Demo

• This work explores utilizing 2D diffusion models for 3D asset generation.
• Improves quality and diversity in 3D shape synthesis.

DiMeR: Disentangled Mesh Reconstruction Model

📄 Paper (arXiv)
Lutao Jiang*, Jiantao Lin*, Kanghao Chen*, Wenhang Ge*, Xin Yang, Yifan Jiang, Yuanhuiyi Lyu, Xu Zheng, Yinchuan Li, Yingcong Chen

Project Page | Code | Demo

• Proposes DiMeR, a geometry-texture disentangled model with 3D supervision for improved sparse-view mesh reconstruction.
• Enhances reconstruction accuracy and efficiency by separating inputs and simplifying mesh extraction.

Advancing high-fidelity 3D and Texture Generation with 2.5D latents

📄 Paper (arXiv)
Xin Yang*, Jiantao Lin*, Yingjie Xu, Haodong Li, Yingcong Chen

• Proposes a unified framework for joint 3D geometry and texture generation using versatile 2.5D representations.
• Improves coherence and quality in text- and image-conditioned 3D generation via 2D foundation models and a lightweight 2.5D-to-3D decoder.

FlexPainter: Flexible and Multi-View Consistent Texture Generation

📄 Paper (arXiv)
Dongyu Yan*, Leyi Wu*, Jiantao Lin, Luozhou Wang, Tianshuo Xu, Zhifei Chen, Zhen Yang, Lie Xu, Shunsi Zhang, Yingcong Chen

Project Page | Code

• Proposes FlexPainter, a multi-modal diffusion-based pipeline for flexible and consistent 3D texture generation.
• Enhances control and coherence by unifying input modalities, synchronizing multi-view generation, and refining textures with 3D-aware models.

PRM: Photometric Stereo-based Large Reconstruction Model

📄 Paper (arXiv)
Wenhang Ge*, Jiantao Lin*, Jiawei Feng, Guibao Shen, Tao Hu, Xinli Xu, Ying-Cong Chen

Project Page | Code | Demo

• A large-scale photometric stereo reconstruction framework.
• Enhances lighting-aware 3D shape recovery with high fidelity.

LucidDreamer: Towards High-Fidelity Text-to-3D Generation via Interval Score Matching

📄 Paper (arXiv)
Yixun Liang*, Xin Yang*, Jiantao Lin, Haodong Li, Xiaogang Xu, Ying-Cong Chen

Code

• Proposes a novel interval score matching approach for text-to-3D generation.
• Significantly improves 3D shape realism and fidelity.

FlexGen: Flexible Multi-View Generation from Text and Image Inputs

📄 Paper (arXiv)
Xinli Xu*, Wenhang Ge*, Jiantao Lin*, Jiawei Feng, Lie Xu, Hanfeng Zhao, Shunsi Zhang, Ying-Cong Chen

Project Page

• A multi-view generation model that fuses text and image inputs.
• Enables controllable and high-quality multi-view synthesis.

SG-Adapter: Enhancing Text-to-Image Generation with Scene Graph Guidance

📄 Paper (arXiv)
Guibao Shen*, Luozhou Wang*, Jiantao Lin, Wenhang Ge, Chaozhe Zhang, Xin Tao, Yuan Zhang, Pengfei Wan, Zhongyuan Wang, Guangyong Chen, Yijun Li, Ying-Cong Chen.

• Introduces scene graph constraints into text-to-image generation.
• Improves structure and semantic consistency in generated images.

Graph Representation and Prototype Learning for Webly Supervised Fine-Grained Image Recognition

📄 Paper (Pattern Recognition Letters)
Jiantao Lin, Tianshui Chen, Ying-Cong Chen, Zhijing Yang, Yuefang Gao

• Proposes a novel graph-based method for fine-grained image recognition under weak supervision.
• Learns better category structures for challenging datasets.

Learning Consistent Global-Local Representation for Cross-Domain Facial Expression Recognition

📄 Paper (ICPR 2022)
Yuhao Xie, Yuefang Gao, Jiantao Lin, Tianshui Chen

Code

• Introduces a domain-adaptive method for cross-domain facial expression recognition.
• Bridges the gap between different facial datasets through global-local feature alignment.