My research lies at the intersection of computer science and arts & design, focusing on generative models and their applications in interdisciplinary fields such as image/video/3D generation and graphic design. Earlier, my research also involved 3D vision.

Selected Publications

Fashionpedia teaser

ART3D: 3D Gaussian Splatting for Text-Guided Artistic Scenes Generation

CVPR The IEEE / CVF Computer Vision and Pattern Recognition Conference, Workshops, 2024.

Pengzhi Li, Chengshuai Tang, Qinxuan Huang, Zhiheng Li

Links: [paper ][poster]
when in doubt teaser

Tuning-Free Image Customization with Image and Text Guidance

ECCV The European Conference on Computer Vision, 2024.

Pengzhi Li, Qiang Nie, Ying Chen, Xi Jiang, Kai Wu, Yuhuan Lin, Jinlong Peng, Yong Liu, Chengjie Wang, Feng Zheng

Links: [project page][paper]
Intentonomy teaser

Generating Daylight-driven Architectural Design from Massing Models via Diffusion Models

CVPR The IEEE / CVF Computer Vision and Pattern Recognition Conference, Workshops, 2024.

Pengzhi Li, Baijuan Li

Links: [project page][pdf]
when in doubt teaser

LayerDiffusion: Layered Controlled Image Editing with Diffusion Models

SIGGRAPH ICML Proceedings of SIGGRAPH Asia 2023, Technical Communications, 2023. (Oral presentation at AI&HCI@ICML 2023)

Pengzhi Li, Qinxuan Huang, Yikang Ding, Zhiheng Li

Links: [project page][pdf] [video]
Intentonomy teaser

Sketch-to-Architecture: Generative AI-aided Architectural Design

PG Pacific Graphics, 2023 (Poster presentation)

Pengzhi Li, Baijuan Li, Zhiheng Li

Daily hottest large model papers on arXiv!

Links: [project page][pdf]
DIFT teaser

Towards Practical Consistent Video Depth Esitimation

ICMR The ACM International Conference on Multimedia Retrieval, 2023 (Oral presentation)

Pengzhi Li, Yikang Ding, Linge Li, Jingwei Guan, Zhiheng Li

Links: [project page][pdf]
VPT teaser

Efficient Temporal Denoising for Improved Depth Map Applications

ICLR The International Conference on Learning Representations, Tiny papers, 2023.

Pengzhi Li, Zhiheng Li

Links: [pdf]
when in doubt teaser

MD2VO: Enhancing Monocular Visual Odometry Through Minimum Depth Difference

IJCNN The International Joint Conference on Neural Networks, 2024.

Pengzhi Li, Chengshuai Tang, Yifu Duan, Zhiheng Li

Links: [pdf]

Research Experience

Fashionpedia teaser

Tencent, Shenzhen, China

Research Intern @ Tencent Youtu Lab

Aug.2023-Jan.2024, Topic: Image Generation and Editing.

Fashionpedia teaser

University Of Pennsylvania, US

Remote Intern @ PennCIS

Apr.2023-Jun.2023, Topic: Machine Learning and stastic.

Fashionpedia teaser

Huawei 2012 Lab, China

Research Intern @ Media Technology Institute

Jun.2022-Dec.2022, Topic: Video Generation and depth estimation.

Fashionpedia teaser

Tsinghua University, China

Master of Electronic Information and Engineering @ ITS Lab

Sep.2021-Jun.2024, Topic: AIGC, 3D vision

Hi, I'm Pengzhi Li, currently a third-year master student majoring in Electronic Information and Engineering at Tsinghua University. My supervisor is Prof. Zhiheng Li. Before joining Tsinghua University, I obtained my bachelor's degree in Architecture from CQJTU. I was fortunately to work as a research intern at Huawei MTI and Tencent YouTu Lab, supervised by Dr. Jingwei Guan and Dr. Qiang Nie.


Currently, my research lies at the intersection of computer science and arts & design, focusing on generative models and their applications in interdisciplinary fields such as image/video/3D generation and graphic design. Earlier, my research also involved 3D vision.


I excel in painting, 3D modeling, and graphic design, and have won awards in several design competitions. Additionally, I am an amateur photographer, specializing in landscape photography. I am looking forward to meeting more outstanding and interesting friends.

I am a passionate polymath with a diverse range of interests, including painting, 3D software modeling with tools like Maya, Blender, UE, and Rhino, computer graphics, graphic design, and architectural design. Captivated by the convergence of these disciplines, I am committed to crafting unique visual masterpieces. I firmly believe that the synergy of art and technology holds the key to boundless potential.

Painting

Fashionpedia teaser

Wuhan University, Hand-drawn sketch

Download: [png]
Fashionpedia teaser

Colosseum, Hand-drawn sketch

Download: [png]
Fashionpedia teaser

Color marker painting

Download: [png]
Fashionpedia teaser

Color marker painting

Download: [png]

Modeling and Animation Rendering

Fashionpedia teaser

3D Modeling and Rendering of the Beijing Bird's Nest and Phoenix Media Center

Download: [video][model]
Fashionpedia teaser

Explosion animation

Download: [video]
Fashionpedia teaser

Heydar Aliyev Center Modeling and Rendering Presentation Animation

Download: [video]
Fashionpedia teaser

Day-night transition animation

Download: [video]

Design

Fashionpedia teaser

Residential Architectural Design

Fashionpedia teaser

Performance Venue Design

Fashionpedia teaser

Space Renovation Design

Fashionpedia teaser

Urban Planning Design

Fashionpedia teaser

Eco-friendly Waterfront Design

Fashionpedia teaser

Historical Building Virtual Reconstruction

Fashionpedia teaser

App Prototype Design

Fashionpedia teaser

Brand Logo Design

I believe that every landscape photo is more than just a visual treat; it's an emotional touchstone. I strive for clarity and vibrancy in my images, but more importantly, I seek to convey the story and emotions behind each scene. Whether you're a photography enthusiast or a traveler with unique insights into natural landscapes, I look forward to exchanging ideas and exploring the boundless possibilities of photography with you.

Photography

Fashionpedia teaser

Ocean Building, Tsinghua University

Download: [png]
Fashionpedia teaser

Nanshan Mountain Park

Download: [png]
Fashionpedia teaser

Shenzhen Talent Park

Download: [png]
Fashionpedia teaser

Shenzhen Bay Bridge

Download: [png]