Haimin Luo | 3D Generation

About

My research centers on 3D generation, with the goal of building controllable, structured, and high-fidelity generative models for visual content creation.

Within this agenda, I use neural rendering as a core modeling tool for geometry, appearance, and editable representations, while digital humans serve as a major application domain spanning hair, humans, motion, and embodied content.

I will obtain PhD degree in Computer Science at ShanghaiTech University in 2026, affiliated with the Visual and Data Intelligence Center (VDI) and the School of Information Science and Technology (SIST). I work with Prof. Jingyi Yu and also collaborate closely with Prof. Lan Xu.

Currently I am also fortunate to collaborate with Prof. Fernando De la Torre as a visiting scholar at the CMU Human Sensing Lab. Previously, I received my bachelor's degree in 2019 from Huazhong University of Science and Technology.

News

[2026.04] HairGPT was conditionally accepted by SIGGRAPH 2026 (journal track).

[2026.02] CGHair was accepted by CVPR 2026. Special thanks to our collaborators at CMU Human Sensing Lab.

[2024.08] Started as a Visiting Scholar at Carnegie Mellon University in Pittsburgh.

[2024.02] GaussianHair: Hair Modeling and Rendering with Light-aware Gaussians was released on arXiv.

[2023.12] MVHuman: Tailoring 2D Diffusion with Multi-view Sampling For Realistic 3D Human Generation was released on arXiv.

[2023.06] NeuralDome was published at CVPR 2023.

[2023.06] Instant-NVR was published at CVPR 2023.

[2023.06] HumanGen was published at CVPR 2023.

[2023.03] NEPHELE: A Neural Platform for Highly Realistic Cloud Radiance Rendering was released on arXiv.

[2022.02] Artemis was published in ACM Transactions on Graphics (Proc. of SIGGRAPH 2022).

[2021.09] Convolutional Neural Opacity Radiance Fields was published at ICCP 2021.

[2021.07] Few-shot Neural Human Performance Rendering from Sparse RGBD Videos was released on arXiv and later published at IJCAI 2021.

[2021] GNeRF was published at ICCV 2021.

Services

Reviewer: SIGGRAPH, SIGGRAPH ASIA, CVPR, ICCV, ECCV, NeurIPS, WACV, AAAI, ACM MM, BMCV, ICLR ...

Research Collaboration: Open to collaborations on 3D generation, especially controllable generation for hair, humans, and motion-rich content, with neural rendering and digital humans as key application domains.

Featured Publications

* denotes equal contribution or advising; † denotes corresponding author

HairGPT: A Unified Autoregressive Framework for 3D Realistic Hairstyle Synthesis

Haimin Luo, Min Ouyang, Lan Xu, Jingyi Yu

April, 2026·SIGGRAPH 2026 (Journal Track, conditional acceptance)

HairGPT is a strand-centric framework that treats authentic strands as generative primitives and formulates 3D hair synthesis as a dual-decoupled autoregressive sequence modeling problem. By decoupling semantic scalp regions and hierarchical strand structure, it enables robust semantic conditioning, compositional editing, and high-fidelity synthesis of rare or complex hairstyles.

Cite

CGHair: Compact Gaussian Hair Reconstruction with Card Clustering

Haimin Luo, Srinjay Sarkar, Albert Mosella-Montoro, Francisco Vicente Carrasco, Fernando De la Torre

February, 2026·CVPR 2026

We present a compact pipeline for hair reconstruction from multi-view images. By clustering hair strands into representative cards and sharing Gaussian texture codebooks, CGHair achieves comparable visual quality with over 200x lower memory footprint and 4x faster strand reconstruction than prior approaches.

Project Page Video Cite

GaussianHair: Hair Modeling and Rendering with Light-aware Gaussians

Haimin Luo, Min Ouyang, Zijun Zhao, Suyi Jiang, Longwen Zhang, Qixuan Zhang, Wei Yang, Lan Xu†, Jingyi Yu†

February, 2024·arXiv 2024

This paper presents GaussianHair, a novel explicit hair representation. It enables comprehensive modeling of hair geometry and appearance from images, fostering innovative illumination effects and dynamic animation capabilities.

Paper Cite Code Dataset Video Project Page

NeuralDome: A Neural Modeling Pipeline on Multi-View Human-Object Interactions

Juze Zhang*, Haimin Luo*, Hongdi Yang, Xinru Xu, Qianyang Wu, Ye Shi, Jingyi Yu, Lan Xu, Jingya Wang

June, 2023·CVPR 2023

We construct a dense multi-view dome to acquire a complex human-object interaction dataset, named HODome, that consists of ~71M frames on 10 subjects interacting with 23 objects. To process the HODome dataset, we develop NeuralDome, a layer-wise neural processing pipeline tailored for accurate tracking, geometry reconstruction, and free-view rendering.

Paper Cite Code Dataset Video Project Page

NEPHELE: A Neural Platform for Highly Realistic Cloud Radiance Rendering

Haimin Luo, Siyuan Zhang, Fuqiang Zhao, Haotian Jing, Penghao Wang, Zhenxiao Yu, Dongxue Yan, Junran Ding, Boyuan Zhang, Qiang Hu, Shu Yin, Lan Xu, Jingyi Yu

March, 2023·arXiv 2023

In this paper, we resort to cloud rendering and present NEPHELE, a neural platform for highly realistic cloud radiance rendering. In stark contrast with existing neural rendering approaches, NEPHELE combines multiple remote GPUs and allows multiple people to view the same NeRF scene simultaneously.

Paper Cite

Artemis: Articulated Neural Pets with Appearance and Motion Synthesis

Haimin Luo, Teng Xu, Yuheng Jiang, Chenglin Zhou, Qiwei Qiu, Yingliang Zhang, Wei Yang, Lan Xu, Jingyi Yu

February, 2022·ACM Transactions on Graphics (Proc. of SIGGRAPH 2022)

We present ARTEMIS, a novel neural modeling and rendering pipeline for generating articulated neural pets with appearance and motion synthesis. ARTEMIS enables interactive motion control, real-time animation, and photo-realistic rendering of furry animals.

Paper Project Page Cite Code Dataset Video

Convolutional Neural Opacity Radiance Fields

Haimin Luo, Anpei Chen, Qixuan Zhang, Bai Pang, Minye Wu, Lan Xu, Jingyi Yu

September, 2021·ICCP 2021

We propose a novel scheme to generate opacity radiance fields with a convolutional neural renderer for fuzzy objects with rich high-frequency details.

Paper Project Page Cite Code Dataset Video

Publications

* denotes equal contribution or advising; † denotes corresponding author

HairGPT: A Unified Autoregressive Framework for 3D Realistic Hairstyle Synthesis

April, 2026·SIGGRAPH 2026 (Journal Track, conditional acceptance)

Haimin Luo, Min Ouyang, Lan Xu, Jingyi Yu

Cite

CGHair: Compact Gaussian Hair Reconstruction with Card Clustering

February, 2026·CVPR 2026

Haimin Luo, Srinjay Sarkar, Albert Mosella-Montoro, Francisco Vicente Carrasco, Fernando De la Torre

Project Video Cite

GaussianHair: Hair Modeling and Rendering with Light-aware Gaussians

arXiv 2024

Haimin Luo, Min Ouyang, Zijun Zhao, Suyi Jiang, Longwen Zhang, Qixuan Zhang, Wei Yang, Lan Xu†, Jingyi Yu†

Project Paper Code Dataset Video Cite

MVHuman: Tailoring 2D Diffusion with Multi-view Sampling For Realistic 3D Human Generation

arXiv 2023

Suyi Jiang, Haimin Luo, Haoran Jiang, Ziyu Wang, Jingyi Yu, Lan Xu

Project Paper Cite

NeuralDome: A Neural Modeling Pipeline on Multi-View Human-Object Interactions

CVPR 2023

Juze Zhang*, Haimin Luo*, Hongdi Yang, Xinru Xu, Qianyang Wu, Ye Shi, Jingyi Yu, Lan Xu, Jingya Wang

Project Paper Code Dataset Video Cite

Instant-NVR: Instant Neural Volumetric Rendering for Human-object Interactions from Monocular RGBD Stream

CVPR 2023

Yuheng Jiang, Kaixin Yao, Zhuo Su, Zehao Shen, Haimin Luo, Lan Xu

Project Paper Video Cite

HumanGen: Generating Human Radiance Fields with Explicit Priors

CVPR 2023

Suyi Jiang, Haoran Jiang, Ziyu Wang, Haimin Luo, Wenzheng Chen, Lan Xu

Project Paper Video Cite

NEPHELE: A Neural Platform for Highly Realistic Cloud Radiance Rendering

arXiv 2023

Haimin Luo, Siyuan Zhang, Fuqiang Zhao, Haotian Jing, Penghao Wang, Zhenxiao Yu, Dongxue Yan, Junran Ding, Boyuan Zhang, Qiang Hu, Shu Yin, Lan Xu, Jingyi Yu

Paper Cite

Artemis: Articulated Neural Pets with Appearance and Motion Synthesis

ACM Transactions on Graphics (Proc. of SIGGRAPH 2022)

Haimin Luo, Teng Xu, Yuheng Jiang, Chenglin Zhou, Qiwei Qiu, Yingliang Zhang, Wei Yang, Lan Xu, Jingyi Yu

Project Paper Code Dataset Video Cite

Few-shot Neural Human Performance Rendering from Sparse RGBD Videos

2021 International Joint Conference on Artificial Intelligence

Anqi Pang, Xin Chen, Haimin Luo, Minye Wu, Jingyi Yu, Lan Xu

Paper Cite

Convolutional Neural Opacity Radiance Fields

2021 IEEE International Conference on Computational Photography

Haimin Luo, Anpei Chen, Qixuan Zhang, Bai Pang, Minye Wu, Lan Xu, Jingyi Yu

Project Paper Code Dataset Video Cite

Gnerf: Gan-based neural radiance field without posed camera

2021 IEEE/CVF International Conference on Computer Vision

Quan Meng, Anpei Chen, Haimin Luo, Minye Wu, Hao Su, Lan Xu, Xuming He, Jingyi Yu

Paper Code Cite

Contact

Feel free to reach out for discussions, collaborations, or visiting opportunities centered on 3D generation, especially projects related to neural rendering, digital humans, hair, and motion-aware content creation.

luohm@shanghaitech.edu.cn

Affiliation

ShanghaiTech University, Visual and Data Intelligence Center (VDI), School of Information Science and Technology (SIST)

Address

393 Middle Huaxia Road, Pudong, Shanghai 201210, China

Profiles

Google Scholar GitHub Twitter CV