GUSH3R: Fast 3D Reconstruction of People and Scenes

🤖 Revolution in 3D Reconstruction: GUSH3R

Researchers from the University of Tokyo have introduced GUSH3R — a framework for the simultaneous reconstruction of dynamic people and static scenes using 3D Gaussian Splatting (3DGS) based on a single monocular video.

🌍 The transition to fast feed-forward models paves the way for creating interactive 3D worlds and real-time digital twins.

👤 It is now possible to create photorealistic 3D scenes with moving people simply by uploading an ordinary video from a single camera.

Source 1: https://abkeito.github.io/gush3r-page/ Source 2: https://arxiv.org/abs/2607.05243