SOLAMI: End-to-End Social Multimodal Model

Immersive Interaction with 3D Autonomous Characters

SOLAMI is the first end-to-end social multimodal model framework for immersive interaction with 3D autonomous characters. Through a unified vision-language-action architecture, automatically generated multimodal interaction dataset, and immersive VR interface, it achieves more natural, precise, and low-latency human-machine social interaction, providing a new solution for building socially intelligent 3D autonomous characters.

Learn More
AnimoXtend

AIGC 3D Generation Plugin

An innovative AIGC 3D generation plugin that brings a new creative experience to Blender users. Through natural language interaction, users can easily generate scenes and character models, and achieve automated character rigging. The plugin includes over 4,000 professional-grade motion capture data and supports flexible retargeting of these animations to any character model. As an all-in-one AI-assisted creation tool, AnimoXtend is dedicated to helping 3D creators efficiently complete the entire workflow from modeling to animation in Blender, significantly improving creation efficiency.

Learn More
Multi-Channel Interaction

Manned Lunar Landing VR Simulation

Originating from the China Astronaut Research and Training Center, this project fully simulates the manned lunar landing mission process through VR equipment integrated with flight joystick simulation and wearable gesture recognition sensors. The system includes multiple functional modules such as an interactive command cabin control panel, simulated augmented reality virtual interface, and remotely operable lunar rover.

Learn More
Node-Based Editing

Gesture-Based Visual Programming System

To improve development and testing efficiency in virtual environments, we developed a VR visual programming tool based on gesture interaction. Addressing the pain point of VR developers needing to frequently switch between real and virtual environments during debugging, we innovatively unified VR software development and debugging within the virtual environment.

Learn More
../../assets/images/about/about.png

Weiye Xiao

Algorithm Researcher -- SenseTime Research
Ph.D. in Engineering -- Southeast University
Former Technical Lead -- SEU Product Reliability Institute
Former Campus Ambassador -- Unity Technology

As a researcher passionate about interdisciplinary technology development, I am dedicated to combining deep learning and computer graphics cutting-edge technologies to explore innovative solutions in content generation. Over the past five years, my research has focused on AIGC, PCG, and 3D digital humans. During my Ph.D. studies, I successfully completed multiple collaborative projects with renowned domestic enterprises, and while working at SenseTime, I contributed to several top-tier conference publications. Currently, my work primarily revolves around multimodal content generation technologies, with a particular focus on large model applications in 3D content creation. I look forward to continuing to delve deep into the intersection of artificial intelligence and computer graphics, contributing to the intelligent evolution of digital content creation.

Learn More
service

Computer Graphics

Five years of experience in graphics development, continuously following cutting-edge computer graphics research. Rich research experience in Rendering, Geometry, and Motion domains. Developed synthetic data generation pipeline that has contributed to multiple outstanding published works.

service

Deep Learning

Over three years of deep learning development experience. Proficient in large model-related knowledge, familiar with common generative model frameworks (GAN, VAE, Diffusion, etc.), and well-versed in popular generative model plugins (LoRA, ControlNet, Redux, etc.). Strong paper reading and implementation capabilities.

service

Design and Art

Professional design and artistic foundation. Proficient in graphic and video software like Figma, PS, FCPX, Davinci, etc. Mastery of 3D creation software including Blender, Maya, Substance Designer, Houdini, etc. Capable of independently completing comprehensive game and film DCC workflows.

Featured Works