Server-side rendering for visualizing volumetric biological data using WebXR

Interactive visualization of volumetric biological datasets in Virtual Reality is currently constrained by the hardware limitations of standalone headsets. Server-side rendering has been proven to address these limitations by offloading computationally intensive rendering tasks to powerful backend servers. WebXR has been previously used for scientific visualization because of its platform-agnostic, browser-based delivery model. We present a server-side rendering pipeline for WebXR visualization of volumetric biological datasets. A GPU-accelerated backend renders stereoscopic 360° equirectangular projections and streams per-eye video to a VR-supported client using WebRTC. The 360° projection enables client-side view-dependent sampling for head rotation, reducing sensitivity to network roundtrips for rotational motion. We evaluate the feasibility of our approach using five open-source volumetric biological datasets, observing significantly higher framerates and the ability to successfully load gigabyte-sized files that fail to render on traditional WebXR

Committee: Oliver Eulenstein (major professor), Amit Kumar Sikder, and Tavis Anderson