Eye contact over video

Abstract

Video communication systems traditionally offer limited or no experience of eye contact due to the offset between cameras and the screen. In response, we are experimenting with the use of multiple Kinect cameras for generating a 3D model of the user, and then rendering a virtual camera angle giving the user an experience of eye contact. In doing this, we use concepts from KinectFusion, such as a volumetric voxel data representation and GPU accelerated ray tracing for viewpoint rendering. This achieves a detailed 3D model from a noisy source, and delivers a promising video output in terms of visual quality, lag and frame rate, enabling the experience of eye contact and face gaze.

Mind Reader

“…As e-textiles advance, better head worn devices could be designed. Humans can notice a difference in angle of five degrees when making eye contact [9]. E-textile head bands could allow for easier transition from the game to direct eye contact.…”

Partially Blended Realities: Aligning Dissimilar Spaces for Distributed Mixed Reality Meetings

“…Collaborative MR systems enable distributed users to meet virtually while retaining presence in their physical space [2,6,16,20,30,31,36]. The interaction metaphor in most of these MR systems is that they bring remote collaborators into the user's physical space by warping them into the local space [18,38].…”

Behaviors, Perceptions, Responsiveness, and Presence: The Dyadic Model of Mediated Communication

“…Much of the HCI research examining eye movement in dyadic VMC has focused on the effects of the technological characteristics of the medium, coming to the conclusion that the camera placement typical of normal computers significantly affected the perception of gaze, even inducing cues associated with lying [4]. Researchers have thus proposed diverse implementations to correct the rendering of the person's eyes on the screen [16,34,37], allowing users to perceive eye contact as in FtF settings. However, human adaptability may render these implementations unnecessary (albeit welcome).…”

“…Although HCI researchers are getting closer to aligning these two components in VMC [16,34,37], the capability of current video-mediated technologies to provide partners with sustained eye-to-eye contact is lacking, which may lead to discrepancies between partners' actual eye movement and their perceptions. Therefore, understanding the effects of both components, and how interactants use technology to compensate for or counteract the effects of reduced gaze, may help designers provide a better user experience that can facilitate presence in video-enabled systems.…”