Yuxiao Yang
I am currently a Master student at Tsinghua University, advised by Prof.Haoqian Wang. Previously, I did my bachelor’s degree in Information Engineering at Zhejiang University.
My research interest includes video generation, 3D generation, and multimodal learning.
📧 Email / 💻 Github / 📚 Google Scholar
💼 Experience
HKUST｜Research Intern｜2024.06 - 2024.11
Baidu｜Research Intern｜2024.12 - 2025.02
Alibaba Cloud｜Research Intern｜2025.02 - 2025.12
MiniMax｜TopTalent Intern｜2025.12 - 2026.01
ByteDance｜Researcher｜2026.02 - Now
📝 Publications
EchoMotion: Unified Human Video and Motion Generation via Dual-Modality Diffusion Transformer
Y Yang, H Sheng, S Cai, J Lin, J Wang, B Deng, J Lu, H Wang, J Ye
🌐 Website 📄 Paper 💻 Code 🤖 Model
- We present EchoMotion, a novel diffusion-based framework for unified human video and motion generation.
Wonder3D++: Cross-Domain Diffusion for High-Fidelity 3D Generation From a Single Image
Y Yang, X Long, Z Dou, C Lin, Y Liu, Q Yan, Y Ma, H Wang, Z Wu, W Yin
📄 Paper 💻 Code 🤖 Model
- We present Wonder3D++, a novel cross-domain diffusion framework for high-fidelity 3D generation from a single image.
Auto-Regressively Generating Multi-View Consistent Images
JK Hu, Y Yang, J Liu, J Wu, C Zhao, Y Lu
📄 Paper 💻 Code 🤖 Model
- We propose a novel auto-regressive model for generating multi-view consistent images.
NOVA3D: Normal Aligned Video Diffusion Model for Single Image to 3D Generation
Y Yang, P Li, Y Zhang, J Lu, X He, M Qin, W Wang, H Wang
📄 Paper
- We introduce NOVA3D, a normal aligned video diffusion model for single image to 3D generation.
Mvreward: Better aligning and evaluating multi-view diffusion models with human preferences
W Wang, H Xu, Y Yang, Z Liu, J Meng, H Wang
📄 Paper
- We propose Mvreward, a novel framework for better aligning and evaluating multi-view diffusion models with human preferences.
Target-Balanced Score Distillation
Z Xu, Q Wang, Y Yang, L Zhang, Z Liang, Y Li
📄 Paper
- We introduce Target-Balanced Score Distillation, a novel method for improving the performance of score-based generative models.