Time ：July 5 ,2016 9:30
Location：Room 1420, Research Building
Hosted by: Division of Fundamental Research Management, VR Lab
Report content: Light fields are image-based representations that use densely sampled rays as a scene description. While the original goal of acquiring a light field is to conduct view synthesis and post-capture refocusing, recent studies have shown that light fields can be extremely useful in other computer vision applications, including stereo matching and 3D reconstruction, stereoscopy synthesis, saliency detection, surveillance, and recognition. In this talk, I present a light field approach for virtual reality (VR) and augmented reality (AR). I first introduce a brief history of VR/AR and the evolution of their core technologies: acquisition, processing, and display. I then discuss the limitation of classical approaches and show why light fields provide a viable path. On the acquisition front, I present several light field capture systems that can produce realistic 3D VR environment beyond regular 360 panoramas. On the processing front, I present several latest 3D reconstruction algorithms that exploit ray geometric structure and sampling patterns of light fields. On the display front, I show how light field head-mounted displays (LF-HMD) can provide unprecedented refocusing capability analogous to human eyes to significantly enhance visual realism. I will summarize my talk by discussing challenges and opportunities of light field based VR/AR approaches.
Speaker CV: Jason is the CTO of Diejing Digital Technologies Ltd. Diejing is a technology and media company focusing on capturing real life experiences for VR and AR. Previously he was head of graphics research and managed performance engineering for GPUs at AMD working on the GPU senior management team and also the office of the CTO. He received his BS and PHD at the Massachusetts Institute of Technology (MIT).