I took a closely look at Steve Seitz and found out his trace of research:
90s: basic theory, analysis and editing of static images– laying the foundation
00-05: tackling with motion:”gaze awareness” “rapid shape acquisition” “Shape recovery for dynamic scenes” “tracking”
06-10: taking the video reconstruction to the next level:” High-resolution capture” –here comes the photosynth: I think it is an exhibit of how static images can be manipulated and reconstructed on a 3D model. The technology experimented in the 90s is mature enough to achieve such function.
11-15: model becomes light-weighted that can be implemented on the mobile phone, of a larger scale( “Megaface: A million faces for recognition at scale”) or real-time (“”Reconstruction and tracking of non-rigid scenes in real-time”), an upgrade to application. The progress of the tech is not focused on creating a VR world, but about analyzing and linking the reality.
16-now: not so many popular publications, focuses come back to creating VR experiences. At the same time using neural network to interact with reality: “Synthesizing obama: learning lip sync from audio”