VISTAlabs

VISTAlabs presents a research preview focused on understanding physical geometry for place verification. Key features include: • Explore room images as depth and 3D point clouds • Generate spatial reconstructions from diverse inputs • Investigate physical geometry for authentication signals • Interactive demos and research notes available VISTAlabs delves into metric 3D understanding to determine if a physical place can serve as an authentication signal. The platform allows users to explore how a scene's observations transform into detailed geometric reconstructions, enabling a deeper understanding of spatial properties. This includes working with RGB, depth, and normal maps that share a pixel grid, alongside 3D point clouds for comprehensive scene analysis. Users can interact with reconstructed environments, such as dragging through a real 180° camera orbit to experience the geometric understanding, rather than just flat image animations. The core functionality centers on taking 360° image inputs and processing them into depth information, camera geometry, metric points, and full 3D representations. The system estimates physical scale, providing endpoint distances directly from retained geometry. VISTAlabs is ideal for researchers, developers, and security professionals interested in exploring novel methods of physical-place verification and spatial biometrics.