State‐of‐the‐Art in GPU‐Based Large‐Scale Volume Visualization

Abstract

This survey gives an overview of the current state of the art in GPU techniques for interactive large‐scale volume visualization. Modern techniques in this field have brought about a sea change in how interactive visualization and analysis of giga‐, tera‐ and petabytes of volume data can be enabled on GPUs. In addition to combining the parallel processing power of GPUs with out‐of‐core methods and data streaming, a major enabler for interactivity is making both the computational and the visualization effort proportional to the amount and resolution of data that is actually visible on screen, i.e. ‘output‐sensitive’ algorithms and system designs. This leads to recent output‐sensitive approaches that are ‘ray‐guided’, ‘visualization‐driven’ or ‘display‐aware’. In this survey, we focus on these characteristics and propose a new categorization of GPU‐based large‐scale volume visualization techniques based on the notions of actual output‐resolution visibility and the current working set of volume bricks—the current subset of data that is minimally required to produce an output image of the desired display resolution. Furthermore, we discuss the differences and similarities of different rendering and data traversal strategies in volume rendering by putting them into a common context—the notion of address translation. For our purposes here, we view parallel (distributed) visualization using clusters as an orthogonal set of techniques that we do not discuss in detail but that can be used in conjunction with what we present in this survey.

A framework for GPU‐accelerated exploration of massive time‐varying rectilinear scalar volumes

“…We briefly discuss here the methods that are most closely related to ours. For a wider coverage, we refer the reader to established surveys on modeling and visualization approaches for time‐varying volumetric data [WF08], compression‐domain DVR [BRGIG∗14], GPU‐based large‐scale DVR [BHP15], and mobile DVR [NJ16].…”

“…They employ multiresolution data representations, compression, out‐of‐core methods and data streaming to enable interactive visualization of massive volumetric datasets. While these architectures have been extremely successful in the exploration of static datasets [TBR∗12,BRGIG∗14, BHP15], current techniques do not fully support real‐time exploration of dynamic data with full spatial and temporal control (see Sec. 2).…”

Parallel Marching Blocks: A Practical Isosurfacing Algorithm for Large Data on Many‐Core Architectures

“…So an accelerating algorithm with high performance and a small memory footprint can offer significant benefits in the GPU-based processing of the large datasets that are becoming the norm [BHP14]; PMB has this potential. The GPUs memory has to accommodate not only the input volume data, but also the output mesh and the accelerating structure involved.…”

A Comparison between GPU-based Volume Ray Casting Implementations: Fragment Shader, Compute Shader, OpenCL, and CUDA

“…With the entry and exit point of one ray into the volume, the problem is commonly reduced to sampling the volume at constant steps, classifying the samples and composing them. It is also accompanied with acceleration techniques like early ray termination [4].…”