AI Insider: Capturing life with AI-enabled volumetric video
AI Insider: Capturing life with AI-enabled volumetric video
Imagine stepping back into a moment, able to move around an object in full 3D. This is the reality of AI-enabled volumetric video, pioneered by researchers at UNSW’s 3DXLab. Multi-camera arrays capture a subject’s full 3D shape and real-time motion in space, transcending flat video by allowing viewers to virtually explore the action from every angle, making the subject a true, interactive 3D model.
The technical leap here is Gaussian splatting, an AI-enhanced process that transforms petabytes of raw capture data into breathtakingly realistic 3D surfaces. Unlike photogrammetry that relies on imperfect meshes, this machine learning method renders millions of blended ellipsoidal shapes. This computational heavy-lifting delivers stunning visual fidelity, handling intricate details like skin texture and fine hair. This innovation creates highly portable data and unlocks applications for advanced training simulations, immersive live events, and the most authentic form of personal memory archiving.
Written by Adam Clark. Adam has spent the past 13 years exploring the world from above by using drones, satellites, and mapping tools to better understand our landscapes. Connect with him on LinkedIn: Adam Clark
Read More: https://www.inside.unsw.edu.au/campus-culture/ai-insider-capturing-life-ai-enabled-volumetric-video
More Recent News: 180 Million Trees Mapped with Lidar in Australia