Guian Fang

Video Generation & World Models

I am a Ph.D. student at Show Lab, National University of Singapore, advised by Prof. Mike Zheng Shou.

My research centers on generative video models and world models for long-form visual generation, currently on few-step distillation, long-horizon memory, and embodied intelligence.

I also work on agents that produce visual content rather than only consume it, and on runtime infrastructure for coding agents. I am open to collaboration.

Previously, I received my B.Eng. in Intelligent Science and Technology at the School of Intelligent Systems Engineering, Sun Yat-sen University, advised by Prof. Xiaodan Liang, co-supervised by Prof. Shengcai Liao.

Selected Work

SolarRenderer

Preprint · 2026

View work

Orbia

Preprint · 2026

View work

Utopai X

Product · 2026 #2 AA T2V

View work

SolarWM

Preprint · 2026

View work

AnyFlow

ECCV 2026 Oral

View work
Claw Orchestrator architecture

Claw Orchestrator

Open Source · 2026

View work

News

  • 2026.10 Released SolarRenderer: turning any structure into finished video in four sampling steps.
  • 2026.10 Released Orbia: a benchmark for whether generated videos form a consistent 3D world, covering 27 world models.
  • 2026.09 Launched Utopai X at Utopai Studios, debuting at #2 on the Artificial Analysis Text to Video leaderboard.

Publications

Representative works are highlighted. * denotes equal contribution. Full list on Google Scholar · CV.

SolarRenderer:
Few-Step Generative Reference Rendering

Guian Fang, Shengju Qian, Dongyue Lu, Junchao Huang, Lutao Jiang, Xin Wang, Yiren Song, Li Jiang, Mike Zheng Shou

Preprint, 2026

Paper Project page Code Weights Dataset
Orbia overview: input preservation, generated-content persistence, and 3D self-consistency, tested with six revisit protocols across 27 models

Orbia:
Do Generated Videos Form a Consistent 3D World?

Dongyue Lu*, Guian Fang*, Hur Lim, Yonggang Wu, Yanrong Wang, Hongshan Chen, Shengju Qian, Yue Liao, Wei Chow, Lingdong Kong, Tianxin Huang, Yu Yang, Hongji Yang, Junchao Huang, Alan Liang, Yihao Wang, Zihan Wang, Rong Li, Hanlin Chen, Xin Wang, Wei Tsang Ooi, Mike Zheng Shou, Shuicheng Yan

Preprint, 2026

Paper Project page Dataset Code

SolarWM H3:
Realtime Interaction

Yihua Du, Junchao Huang, Bohong Chen, Yaoyang Liu, Zixin Zhang, Xianghao Kong, Dongyu Yan, Zhuoran Zhao, Guian Fang, Yajing Chen, Shengju Qian

Blog post, 2026

Blog

H3 Streaming VAE Decoder:
Training-Free Streaming without Future Latents

Bohong Chen, Yajing Chen, Yihua Du, Junchao Huang, Guian Fang, Shengju Qian

Blog post, 2026

Blog Code
Mask Forcing pipeline — masking the self-rollout with two noise levels widens the student's coverage of the teacher distribution

Mask Forcing:
Improving Autoregressive Video Diffusion Distillation via Dual-Noise Masking Rollout

Zhuoran Zhao, Shengju Qian, Tongtong Liang, Xianghao Kong, Songchun Zhang, Junchao Huang, Guian Fang, Xin Wang, Pan Hui, Anyi Rao

Preprint, 2026

Paper Project page Code

PAI-Actor:
Cinematic Multi-Character Replacement in Dynamic Scenes

Bangxun Tang*, Heyuan Gao*, Yiren Song*, Guian Fang, Zijian He, Jie Yang, Mike Zheng Shou

NeurIPS, 2026

Paper Project page Code
SolarWM teaser — one framework generating diverse long-horizon interactive worlds

SolarWM:
Open Data and Scalable Training for Long-Horizon Video World Models

Junchao Huang*, Guian Fang*, Shengju Qian, Xianghao Kong, Zhuoran Zhao, Wei Huang, Yihua Du, Zixin Zhang, Justin Cui, Yuchao Gu, Yukang Chen, Xinting Hu, Tianyu He, Shaoshuai Shi, Zhuotao Tian, Xin Wang, Mike Zheng Shou, Li Jiang

Preprint, 2026

Paper Project page Models Dataset Code

AnyFlow:
Any-Step Video Diffusion Model with On-Policy Flow Map Distillation

Yuchao Gu, Guian Fang, Yuxin Jiang, Weijia Mao, Song Han, Han Cai, Mike Zheng Shou

ECCV, 2026Oral

Merged into diffusers FastVideo FastGen

Paper Project page Models

PAI-Studio:
Cinematic Video Background Replacement with Camera-Aware Motion

Heyuan Gao*, Bangxun Tang*, Yiren Song*, Guian Fang, Zijian He, Jie Yang, Mike Zheng Shou

ECCV, 2026

Paper Project page Code
Declare, Compile, Look pipeline — plan a typed FigureSpec, compile the geometry deterministically, emit figures, and repair by patching the spec

Declare, Compile, Look:
Coordinate-Free Layout Generation with Vision-in-the-Loop Repair

Guian Fang, Mengsha Liu, Mike Zheng Shou

ECCV Workshop (MDA), 2026Talk

Paper
VEditBench overview — 420 real-world videos, 6 editing tasks, 9 evaluation dimensions

VEditBench:
Holistic Benchmark for Text-Guided Video Editing

Jay Zhangjie Wu*, Guian Fang*, Dongrong Joe Fu, Vijay Anand Raghava Kanakagiri, Forrest Iandola, Kurt Keutzer, Wynne Hsu, Zhen Dong, Mike Zheng Shou

Preprint, 2025

Paper
HumanRefiner research visualization showing human pose refinement

HumanRefiner:
Benchmarking Abnormal Human Generation and Refining with Coarse-to-fine Pose-Reversible Guidance

Guian Fang*, Wenbiao Yan*, Yuanfan Guo*, Jianhua Han, Zutao Jiang, Hang Xu, Shengcai Liao, Xiaodan Liang

ECCV, 2024

Dataset page Paper
T2VScore framework — text alignment and video quality evaluation pipelines

T2VScore:
Towards A Better Metric for Text-to-Video Generation

Jay Zhangjie Wu*, Guian Fang*, Haoning Wu*, Xintao Wang, Yixiao Ge, Xiaodong Cun, David Junhao Zhang, Jia-Wei Liu, Yuchao Gu, Rui Zhao, Weisi Lin, Wynne Hsu, Ying Shan, Mike Zheng Shou

Preprint, 2024

Paper Project page Code
ChartThinker framework diagram showing contextual chain-of-thought approach

ChartThinker:
A Contextual Chain-of-Thought Approach to Optimized Chart Summarization

Mengsha Liu, Daoyuan Chen, Yaliang Li, Guian Fang, Ying Shen

LREC-COLING, 2024

Dataset page Paper
RealignDiff framework showing coarse-to-fine semantic re-alignment process

RealignDiff:
Boosting Text-to-Image Diffusion Model with Coarse-to-fine Semantic Re-alignment

Guian Fang*, Zutao Jiang*, Jianhua Han, Guansong Lu, Hang Xu, Shengcai Liao, Xiaojun Chang, Xiaodan Liang

IEEE TNNLS, 2023

Paper

Products & Open Source

Deployed products and open-source systems I've led or co-built — productized research rather than peer-reviewed papers.

Collaborations

The teams I've worked with, and my closest collaborators at each.

Lightspeed Studios, Tencent

@Shengju Qian
@Zeyu Hu
@Xin Wang

Utopai Studios

@Jie Yang
@Zijian He

MikoAI Ltd.

@Yuan Ma
@Zhaosheng Chi
@Yunyi Zhu

OpenGVLab, Shanghai AI Lab

@Peng Gao
@Dongyang Liu

Noah's Ark Lab, Huawei

@Hang Xu
@Jianhua Han

NVIDIA Research

@Yuchao Gu
@Song Han
@Han Cai

Activities & Services

Conference Reviewer

  • CVECCV, CVPR, ICCV
  • MLNeurIPS, ICLR, ICML
  • AIAAAI, AISTATS
  • NLPACL Rolling Review (ACL, EMNLP, NAACL, EACL)

Workshop Organizer

  • 2024LOVEU Workshop @ CVPR 2024: Long-form Video Understanding Towards Multimodal AI Assistant and Copilot

Teaching Assistant

  • EE3703Machine Learning with Applications
  • EE4309Robot Perception
  • EE5106Advanced Robotics
  • EE6733Advanced Topics on Vision and Machine Learning

Honors & Awards