Stereo frame decomposition for error-constrained remote visualization

As growth in dataset sizes continues to exceed growth in available bandwidth, new solutions are needed to facilitate efficient visual analysis workflows. Remote visualization can enable the colocation of visual analysis compute resources with simulation compute resources, reducing the impact of bandwidth constraints. While there are many off-the-shelf solutions available for general remoting needs, there is substantial room for improvement in the interactivity they offer, and none focus on supporting stereo remote visualization with programmable error bounds. We propose a novel system enabling efficient compression of stereo video streams using standard codecs that can be integrated with existing remoting solutions, while at the same time offering error constraints that provide users with fidelity guarantees. By taking advantage of interocular coherence, the flexibility permitted by error constraints, and knowledge of scene depth and camera information, our system offers improved remote visualization frame rates.