Research

Abstract

Sample based ray marching is an effective method for direct volume rendering of unstructured meshes. However, sampling such meshes remains expensive, and strategies to reduce the number of samples taken have received relatively little attention. In this paper, we introduce a method for rendering unstructured meshes using a combination of a coarse spatial acceleration structure and hardware-accelerated ray tracing. Our approach enables efficient empty space skipping and adaptive sampling of unstructured meshes, and outperforms a reference ray marcher by up to 7x.

Abstract

We explore a first proof-of-concept example of creatively using the Turing generation's hardware ray tracing cores to solve a problem other than classical ray tracing, specifically, point location in unstructured tetraheral meshes. Starting with a CUDA reference method, we describe and evaluate three different approaches to reformulate this problem in a manner that allows it to be mapped to these new hardware units. Each variant replaces the simpler problem of point queries with the more complex one of ray queries; however, thanks to hardware acceleration, these approaches are actually faster than the reference method.

Abstract

We present a parallel quadtree algorithm that resolves between geometric objects, modeling space between objects rather than the objects themselves. Our quadtree has the property that no cell intersects more than one labeled object. A popular technique for discretizing space is to impose a uniform grid—an approach that is easily parallelizable but often fails because object separation is not known a priori or because the number of cells required to resolve closely spaced objects exceeds available memory. Previous parallel algorithms that are spatially adaptive, i.e., discretizing finely only where needed, either separate points only or make no guarantees of object separation. Our 2D algorithm is the first to construct an object-resolving discretization that is hierarchical (saving memory) yet with a fully parallel approach (saving time). We describe our algorithm, demonstrate experimental results, and discuss extension to 3D. Our results show significant improvement over the current state of the art.

Abstract

This paper presents an open source implementation of a volume visualizer capable of rendering large scale simulated and tomographic volume datasets on everyday commodity hardware. The ability to visualize this data enables a wide range of analysis in many scientific disciplines, such as medicine, biology, engineering, and physics. The implementation presented takes advantage of a hierarchical space filling curve to efficiently query only a subset of the data required for accurate visualization. A major advantage of this implementation is that it is open source, and works on a wide range of visualization devices by using the Unity game engine, a cross platform engine capable of targeting over 25 platforms. This work's unique contribution is utilizing a space filling curve for multiple bricks in the volume to permit adaptive streaming at different resolutions.

More coming soon