Real-Time Rendering of Dynamic Line Sets using Voxel Ray Tracing

TL;DR

This work tackles real-time rendering of large, dynamic line sets with global illumination and accurate transparency. It introduces a GPU-based pipeline that voxelizes lines into a conservative capsule-based occupancy pyramid each frame, applies camera-visible culling, and constructs a two-pass A-Buffer in voxel space for efficient shading. Shading is performed in voxel space using voxel cone tracing, followed by a voxel-space ray-tracing pass that achieves ground-truth transparency without per-frame precomputation. The approach offers strong performance advantages for opaque dynamic line sets and produces high-quality semi-opaque renderings, with potential extensions to level-of-detail and interactive tractography applications.

Abstract

Real-time rendering of dynamic line sets is relevant in many visualization tasks, including unsteady flow visualization and interactive white matter reconstruction from Magnetic Resonance Imaging. High-quality global illumination and transparency are important for conveying the spatial structure of dense line sets, yet remain difficult to achieve at interactive rates. We propose an efficient voxel-based ray-tracing framework for rendering large dynamic line sets with ambient occlusion and ground-truth transparency. The framework introduces a voxelization algorithm that supports efficient construction of acceleration structures for both voxel cone tracing and ray tracing. To further reduce per-frame preprocessing cost, we developed a voxel-based culling method that restricts acceleration structure construction to camera-visible voxels. Together, these contributions enable high-quality, real-time rendering of large-scale dynamic line sets with physically accurate transparency. The results show that our method outperforms the state of the art in quality and performance when rendering (semi-)opaque dynamic line sets.

Figures (10)

• Figure 1: Schematic overview of the rendering pipeline and its intermediate outputs. Each step in the pipeline, represented by blue blocks, is described in sections \ref{['method:voxelization']}-\ref{['method:rendering']}. Green blocks represent (intermediate) data structures, while arrows show the in/outputs of each pipeline step.
• Figure 2: Illustration of our voxel traversal scheme, shown in 2D. The algorithm traverses axis $a_0$, visiting points $\mathbf{p_0}$-$\mathbf{p_6}$. For every consecutive pair of points a bounding box (dark blue) is computed along the minor axes $a_1, a_2$ using the projected radii $\hat{r}_1, \hat{r}_2$. All intersected voxels (light blue) are voxelized. See the text for details.
• Figure 3: Illustration of the SDF-based occupancy estimation, shown in 2D. To compute the occupancy for a given capsule-voxel intersection, the clipped capsule occupancy function (Algorithm \ref{['algorithm:capsule:occupancy']}, shown as the blue gradient) is sampled at the center of each intersecting voxel (filled white-blue circles). See text for more details.
• Figure 4: Volume renderings of the occupancy pyramid of the Aneurysm line set. Without clipping, spherical artifacts appear at every interior vertex. When using the clipped capsule SDF, artifacts are resolved and aliasing is significantly reduced compared to DDA voxelization.
• Figure 5: Volume renderings of the occupancy pyramid of the Arcuate Fasciculus line set. Without Phone-Wire Anti-Aliasing, thin lines introduce aliasing and an overestimation of total occupancy. With Phone-Wire Anti-Aliasing these artifacts are reduced.