Yizhi Wang

I am currently a Senior Research Scientist at ByteDance/TikTok, working on video synthesis and generation.

Previously, I was a Postdoc Fellow supervised by Professor Hao (Richard) Zhang in GrUVi Lab at Simon Fraser University (SFU), Canada. During my postdoc, I worked on 3D shape reconstruction and content creation.

I received my PhD degree from Peking University, where I worked on Computer Vision and Computer Graphics. During my PhD, I focused on applying deep generative models to analyze and synthesize 2D geometric data such as glyphs, fonts, and layouts. I was supervised by Prof. Zhouhui Lian and Prof. Jianguo Xiao.

News

Oct 2026SplitMoE is accepted by NeurIPS 2026 as a spotlight paper!
Feb 2026VIVA is accepted by CVPR 2026.
Jan 2026MAGREF is accepted by ICLR 2026.
Nov 2025ACT-R is accepted by 3DV 2026.
Jun 2025Selected as an Outstanding Reviewer for CVPR 2025.
Jan 2025GALA is accepted by ICLR 2025.
Jul 2024SweepNet is accepted by ECCV 2024.
May 2024EASI-Tex is accepted by SIGGRAPH 2024 as a Journal Paper.
Mar 2024Slice3D is accepted by CVPR 2024. Code has been released.
Jan 2024I am co-organizing the Workshop on Graphic Design Understanding and Generation in CVPR 2024.
Dec 2023Introducing Slice3D, a new notion for single-view 3D reconstruction via multi-slice reasoning.
Jul 2023DS-Fusion is accepted by ICCV 2023. Check out the fun video.
Jun 2023Check how Christina uses DeepVecFont in her art-design project (Video and Blog).

Representative papers are highlighted.

Breaking the Uniformity Trap: Scaling Video Diffusion Model via SplitMoE

Yu Xu, Yuxin Zhang, Xiao Yang, Haotian Yang, Yizhi Wang, Xinwei Huang, Minxuan Lin, Angtian Wang, Chongyang Ma, Fan Tang

NeurIPS 2026 Spotlight

Project Page / arXiv

A split-role sparse architecture for video diffusion models that enables semantic-aware expert routing without uniform load-balancing constraints.
PDMD: Projected Distribution Matching Distillation for Video Diffusion Models

Zimo Wang, Junkun Yuan, Angtian Wang, Haotian Yang, Canyu Zhang, Siyuan Yuan, Xingchang Huang, Bo Liu, Yizhi Wang, Yiding Yang, Chongyang Ma, Gordon Guocheng Qian

arXiv 2026

Project Page / arXiv / Code / Models

A one-line projection that filters critic error from distribution-matching updates, stabilizing few-step video and joint audio-video diffusion distillation without extra losses or networks.
DMAD: Distribution Matching as Adversarial Distillation for Fast Visual Generation

Zhengming Yu, Junkun Yuan, Haotian Yang, Gordon Guocheng Qian, Yizhi Wang, Angtian Wang, Yiding Yang, Bo Liu, Xin Li, Wenping Wang, Chongyang Ma

arXiv 2026

Project Page / arXiv / Code / Models / Video

An adversarial formulation of distribution matching that learns log-density ratios with two discriminator heads for fast image, video, and joint audio-video generation.
VIVA: VLM-Guided Instruction-Based Video Editing with Reward Optimization

Xiaoyan Cong, Haotian Yang, Angtian Wang, Yizhi Wang, Yiding Yang, Canyu Zhang, Chongyang Ma

CVPR 2026

Project Page / arXiv / Code

A video editing framework that encodes editing instructions with a vision-language model and leverages GRPO to enhance editing performance.
MAGREF: Masked Guidance for Any-Reference Video Generation with Subject Disentanglement

Yufan Deng, Yuanyang Yin, Xun Guo, Yizhi Wang, Jacob Zhiyuan Fang, Shenghai Yuan, Yiding Yang, Angtian Wang, Bo Liu, Haibin Huang, Chongyang Ma

ICLR 2026

Project Page / arXiv / Code / Video

A novel conditioning approach with subject disentanglement for identity-consistent video generation from multiple reference images.
GALA: Geometry-Aware Local Adaptive Grids for Detailed 3D Generation

Dingdong Yang, Yizhi Wang, Konrad Schindler, Ali Mahdavi-Amiri, Hao (Richard) Zhang

ICLR 2025

Project Page / arXiv / Code / Video

A novel representation of 3D shapes using oriented and anisotropic local grids.
Slice3D: Multi-Slice, Occlusion-Revealing, Single View 3D Reconstruction

Yizhi Wang, Wallace Lira, Wenqi Wang, Ali Mahdavi-Amiri, Hao (Richard) Zhang

CVPR 2024

Project Page / arXiv / Code / Video

Slice3D predicts multi-slice images to reveal occluded parts without changing the camera, then lifts the slices into a 3D model.
SweepNet: Unsupervised Learning Shape Abstraction via Neural Sweepers

Mingrui Zhao, Yizhi Wang, Fenggen Yu, Changqing Zou, Ali Mahdavi-Amiri

ECCV 2024

Project Page / arXiv / Code / Video

A novel approach for shape abstraction via sweep surfaces, using superellipses for profiles and B-spline curves for axes.
EASI-Tex: Edge-Aware Mesh Texturing from Single Image

Sai Raj Kishore Perla, Yizhi Wang, Ali Mahdavi-Amiri, Hao (Richard) Zhang

ACM Transactions on Graphics (SIGGRAPH Journal-Track) 2024

Project Page / arXiv / Code / Video

A single-image mesh texturing approach that employs diffusion models with judicious conditioning to seamlessly transfer textures.
ARO-Net: Learning Implicit Fields from Anchored Radial Observations

Yizhi Wang*, Zeyu Huang*, Ariel Shamir, Hui Huang, Hao (Richard) Zhang, Ruizhen Hu

CVPR 2023

Project Page / arXiv / Code / Video

A category-agnostic shape encoding for learning neural field representations amid significant shape variations.
DS-Fusion: Artistic Typography via Discriminated and Stylized Diffusion

Maham Tanveer, Yizhi Wang, Ali Mahdavi-Amiri, Hao (Richard) Zhang

ICCV 2023

Project Page / arXiv / Code

Automatically generate artistic typography by stylizing letter fonts to convey the semantics of an input word.
BRICS: Bi-level feature Representation of Image CollectionS

Dingdong Yang, Yizhi Wang, Ali Mahdavi-Amiri, Hao (Richard) Zhang

arXiv 2023

Project Page / arXiv / Code

A bi-level feature representation for an image collection, with a per-image latent space above a multi-scale feature grid space.
DeepVecFont-v2: Exploiting Transformers to Synthesize Vector Fonts with Higher Quality

Yuqing Wang, Yizhi Wang, Longhui Yu, Yuesheng Zhu, Zhouhui Lian

CVPR 2023

Project Page / arXiv / Code

Employing Transformers and a relaxed representation for higher-quality vector font generation, including Chinese glyphs.
Aesthetic Text Logo Synthesis via Content-aware Layout Inferring

Yizhi Wang, Guo Pu, Wenhan Luo, Yexin Wang, Pengfei Xiong, Hongwen Kang, Zhouhui Lian

CVPR 2022

Project Page / arXiv / Code

A content-aware layout generation network that synthesizes aesthetic layouts from element images and their content.
DeepVecFont: Synthesizing High-quality Vector Fonts via Dual-modality Learning

Yizhi Wang, Zhouhui Lian

ACM Transactions on Graphics (SIGGRAPH Asia 2021 Technical Paper) 2021

Project Page / arXiv / Code

Synthesizing vector fonts via dual-modality learning and differentiable rasterization instead of rule-based vectorization.
Attribute2Font: Creating Fonts You Want From Attributes

Yizhi Wang*, Yue Gao*, Zhouhui Lian

ACM Transactions on Graphics (SIGGRAPH 2020 Technical Paper) 2020

Project Page / arXiv / Code

Synthesizing fonts according to user-specified attributes such as italic, serif, cursive, and angularity.
Exploring Font-independent Features for Scene Text Recognition

Yizhi Wang, Zhouhui Lian

ACM Multimedia 2020

Project Page / arXiv / Code

A font-independent feature representation method to improve scene text recognition robustness under font variance.
Font Recognition in Natural Images via Transfer Learning

Yizhi Wang, Zhouhui Lian

MMM 2018

Project Page / Paper / Dataset

Recognizing font styles of texts in natural images with deep learning and transfer learning.

Experience

Mar. 2026 - Now
Senior Research Scientist

ByteDance / TikTok, USA
Nov. 2024 - Mar. 2026
Research Scientist

ByteDance / TikTok, USA
Jan. 2023 - Nov. 2024
Postdoctoral Fellow

Simon Fraser University, Canada

Supervisor: Prof. Richard (Hao) Zhang
Jun. 2021 - Nov. 2021
Research Intern

Tencent

Advisor: Dr. Wenhan Luo

Service

Journal Reviews

IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)
IEEE Transactions on Visualization and Computer Graphics (TVCG)
Expert Systems with Applications

Conference Reviews

SIGGRAPH 2026
SIGGRAPH Asia 2024, 2025
CVPR 2022, 2023, 2024, 2025, 2026
ICCV 2021, 2023, 2025
ECCV 2022, 2024
AAAI 2022, 2023, 2024
Outstanding Reviewer, CVPR 2025

Yizhi Wang (王逸之)

News

Experience

Service

Journal Reviews

Conference Reviews