Memory‐Efficient GPU Volume Path Tracing of AMR Data Using the Dual Mesh

Abstract

Davis 3 NVIDIA Figure 1: Overview of our method. Given a block-structured or octree-AMR data set (left) we first create the dual mesh (middle) and split that into the truly unstructured elements used to stitch the level boundaries (red) and those that are regular voxels (blue/white checkered). We then cluster voxels to become "gridlets". Right: gridlets colorized by their ID. We build a bounding volume hierarchy over the gridlets and the remaining unstructured elements. The result is a sampleable representation that generates the exact same result as sampling on the dual mesh directly, but with significantly lower memory overhead and higher sampling speed. On our largest data sets, we see memory savings of up to 3× compared to highly compressed state-of-the-art unstructured mesh representations.

Beyond ExaBricks: GPU Volume Path Tracing of AMR Data

“…The ExaBricks framework and data structure have been explained in detail by prior works [WZU*21,ZSM*22,ZWS*22]. We refer the reader to these for details.…”

“…For rendering octree‐based AMR specifically, Wang et al [WMU*20] developed high‐quality reconstruction filters for direct ray tracing. Other authors have also focused on large‐scale out‐of‐core AMR rendering, such as the interactive streaming and caching framework proposed by Wu et al [WDM22] targeted at CPU rendering, or the framework by Zellmann et al [ZWS*22], who proposed a similar method for AMR streaming and rendering on GPUs, but focused on time‐varying data.…”

“…GPU path tracing native AMR data, to our knowledge, has not been proposed by the literature yet. Closest to this are OpenVKL [KJM21] that can path trace on the CPU but does not use any of the optimizations necessary for GPUs, and the framework proposed by Zellmann et al [ZWMW23] that renders the AMR data by converting it to an unstructured mesh with voxels.…”