Generation of Intermediate Viewpoints for Scalable Adaptation of Real World Environments for Virtual Reality

Virtual Reality (VR) is a rapidly growing field with innovative applications. Two main forms of VR content can be identified in current applications: Computer generated 3D environments and adaptations of real world scenarios through means of digital imagery and video. While the latter provides a more realistic experience, scalable content generation methods for adaptations of real world environments to virtual space is still unavailable. Needless to say achieving unrestricted navigability in such environments, on a mobile platform with near real time response is quite challenging. The paper presents the algorithm for scalable intermediate view generation, devised as a part of a research to develop an universal content generation platform for real world environments in virtual reality. The algorithm uses adjacent equirectangular images to recreate the 360 degrees x 180 degrees view of any point in-between, taking directional deformation into account. Input can be equirectangular images captured using any device, including 360 cameras and panoramas captured through mobile phones. The output can be rendered on specialized VR headsets or mobile phones. The algorithm is implemented with mobile platform as the target, thus rendering is optimized for limited resource utilization. The paper presents the concept, design and optimization used in this method.