Technology
Innovation Hub
Explore our cutting-edge technology ecosystem, from 3D Gaussian Splatting to WebXR platforms, from RAG digital humans to AI lip-sync animation. We are pioneering the future of digital experiences.
Our Stack
Transforms real-world spaces into digital assets through 3D reconstruction.
Brings memories to life with AI-powered understanding and generation.
Delivers immersive experiences through VR and web platforms.
01 Capture/Create
Scanning & Digitization
Transforming real-world spaces into digital assets through advanced 3D reconstruction and spatial capture technologies.
3DGS (3D Gaussian Splatting)
Fast, photorealistic 3D capture using standard cameras.
Advanced method for reconstructing real-world environments into high-fidelity 3D scenes using standard RGB cameras. Students visit rural homes and historical locations to film elderly participants, creating immersive, photorealistic virtual environments that preserve personal memories and cultural spaces in VR, web, and LED installations.
MemoLens
AR-guided 360° recording for precise spatial capture.
An iOS companion app that transforms everyday phone videos into the foundation for immersive 3D spaces. By utilizing AR-guided shooting, MemoLens ensures the strict spatial accuracy required for high-quality 3D Gaussian Splatting reconstruction. The capture process is entirely offline, guaranteeing that your everyday videos and captured materials remain fully under your control before the 3D generation process.
Autonomous LiDAR Scanner
Autonomous LiDAR robot for spatial data collection.
An autonomous four-legged robot (Unitree Go2) with dual Livox Mid-360 LiDAR sensors and onboard edge computing for real-time 3D SLAM. It delivers physically accurate trajectories in textureless environments, producing high-fidelity point clouds that accelerate 3D Gaussian Splatting convergence. Traverses stairs and uneven terrain to create georeferenced digital twin assets beyond manual scanning reach.
02 Edit
AI-Powered Generation
Leveraging cutting-edge AI models to generate, enhance, and animate digital content with human-like intelligence.
AIGC Scene Creation
Text to 3D — transform imagination into immersive scenes.
Leveraging cutting-edge AI models to generate high-fidelity 3D Gaussian Splatting environments from simple text prompts, bridging the gap between abstract concepts and digital reality. Our restoration tool utilizes AIGC to reconstruct lost sites and moments into immersive 3DGS scenes. By interpreting text descriptions and archival data, it allows you to explore the past with photorealistic accuracy, preserving cultural heritage and turning ancient narratives into tangible, 3D digital experiences.
LLM & RAG Digital Humans
Adaptive conversations tailored to audience and context.
Large Language Models with Retrieval-Augmented Generation pull context-specific information from curated memory bases of elderly interviews and historical data. RAG-powered digital humans adapt tone, content, and emotional depth based on users, ensuring accuracy, emotional resonance, and ethical storytelling for deep, reflective interactions.
AI LipSync Animation
Realistic facial motion synced with voice.
AI-driven facial animation technology that matches spoken audio to realistic mouth movements and facial expressions on digital avatars. After voice cloning and translation, LipSync animates avatars faces in real-time, enhancing emotional connection and realism, especially crucial for war veterans and elderly speakers storytelling.
03 Experience
Immersive Visualization
Delivering unforgettable immersive experiences across VR, web, and location-based entertainment platforms.
MemoSpace
Live inside your memories in immersive VR.
A dedicated VisionOS application designed to let users experience reconstructed 3D scenes in virtual reality. By importing environments generated via open-source 3DGS engines, MemoSpace enables users to walk back into preserved personal memories and cultural spaces. It serves as a seamless bridge between memory and immersion, allowing families and communities to revisit stories together while ensuring all data remains safely on local devices.
VR LBE Experience
Group exploration of 3D scenes in shared space.
HTC Location-Based Entertainment technology transforms spaces into shared virtual learning environments. Supporting up to 20 participants within 200-square-meter areas, users simultaneously interact with AI-powered digital humans and explore 3D reconstructed scenes, fostering collective empathy and dialogue through deeply social, embodied learning experiences.
VR LBE Experience
10 people within 100m²
WebXR Platform
Immersive VR scenes directly in your web browser.
Self-developed platform built on WebXR technology, enabling seamless access to immersive 3DGS scenes through web browsers—no installation needed. Presents reconstructed homes of marginalized elderly alongside AI-powered digital humans, removing technical barriers and promoting inclusive, globally accessible empathy learning.
Interactive LED Experience Room
Multi-sensory storytelling with motion, sound, and light.
Custom-built immersive LED room designed like a traditional riverboat, integrating high-resolution panoramic LED walls, floor projections simulating flowing water, and vibration-enabled seats. AI-powered digital humans share memories of life in Hangzhou and Yunnan, creating a bridge of empathy between generations through culturally rich, multi-sensory experiences.
AIGC Crystal Balls
Preserving faces, voices, and stories in time.
Using cutting-edge AIGC technology, we craft lifelike video portraits of elderly participants, capturing their facial expressions, voices, and personal narratives. These portraits are elegantly housed within crystal ball installations as enduring keepsakes — tangible, beautiful, and deeply personal. Future generations can interact with these digital likenesses, experiencing the warmth and wisdom of their ancestors in a format that is both engaging and accessible.
