A Framework for Compact and Improved Panoramic VR Dissemination

Abstract

Panoramic capture devices in Cultural Heritage are becoming widely available to consumer market, also due to comfortable interactive online dissemination and to the growth of VR segment. VR fruition through an HMD although, requires a virtual 3D representation to provide consistency in terms of experience, scale and spatial perception, overcoming limitations of standard approaches in orientation+positional HMD tracking model. However, modeling of 3D scenes and especially optimization of acquired dataset, are often time-consuming tasks: these are further stressed when dealing with latency-free demands of latest HMDs. In this paper, we propose a novel framework for panoramic acquisition and an improved data model for VR dissemination: spherical panoramas, omnidirectional depth-maps and semantic annotations are encoded into a compact, coherent representation that suits modern HMD needs and low-cost VR devices. We describe advantages of our approach in terms of acquisition pipeline, presence and depth perception in HMDs fruition, discussing also visualization efficiency in online contexts. We present a few case studies where we applied the methodology and the workflow we adopted, comparing results. We discuss integration of existing desktop toolkits into the pipeline, dissemination capabilities through recent WebVR API and framework advantages for immersive VR panoramic video streaming.