Ferramentas de edição e visualização para vídeos 360º interativos

Abstract

Video editing is essential for collecting and arranging multimedia content to fit a consistent publication, especially prominent in journalistic venues. Orthogonally, Virtual Reality (VR) experiences evoke in their users a feeling known as presence: multi-sensory assimilation of a Virtual Environment (VE). Combining this with established news video editing techniques, one arrives at the concept of immersive journalism. One of the challenges of immersive journalism is 360º video editing - when the focus is no longer restricted, conditioning users' attention must be done by different means, namely virtual indicators. However, the distortion applied to the imagery makes adding special effects non-trivial. Such effects, called annotations due to their informational nature, are spatialised two or three-dimensional objects that are displayed as an overlay. To solve this issue, a video player capable of rendering these effects in real time is proposed. The purpose of this dissertation is to create, evaluate and compare a set of VE-interaction techniques that enable a set of authoring operations within a VR 360º video editor. As such, a prototype was developed with the Unity engine to provide a testing environment. It allows users to annotate 360º video using keyframing mechanisms. First an annotation paradigm was developed, with each annotation being written in HTML and rendered onto a plane in 3D space. Second, interaction metaphors based on pointing, clicking and grabbing gestures were implemented, with slight modifications for a selected set of input devices. Standard mouse implementation utilises pointing, clicking and dragging using mouse clicks. The HTC Vive approach employs the proprietary controller-wands like a laser pointing device, with the trigger button being used to indicate selection or dragging. Finally, the Leap Motion controller provided a more natural interface in which users can use their hands to point and use a pinching gesture to trigger selecting and dragging. Furthermore, within this scope additional questions arise: is a VE the most adequate for 360º video editing? Which data storage techniques are most appropriate to ensure compatibility among potential future implementations? In brief, which interaction techniques are more adequate for use in 360º video editing, in terms of efficiency, fatigue and naturalness? To assess this, a usability test was conducted requiring users to perform annotating tasks of increasing complexity while telemetry was gathered by the system. Additional questionnaires were handed out to score each interaction technique in a Usefulness, Satisfaction, and Ease of use (USE) test scale. Currently the implemented prototype provides an environment that allows users to apply the researched techniques to the annotating of 360º video effectively.