Kelly P. Gaither

Illustration and photography inspired visualization of flows and volumes

Understanding and analyzing complex volumetrically varying data is a difficult problem. Many computational visualization techniques have had only limited success in succinctly portraying the structure of three-dimensional turbulent flow. Motivated by both the extensive history and success of illustration and photographic flow visualization techniques, we have developed a new interactive volume rendering and visualization system for flows and volumes that simulates and enhances traditional illustration, experimental advection, and photographic flow visualization techniques. Our system uses a combination of varying focal and contextual illustrative styles, new advanced two-dimensional transfer functions, enhanced Schlieren and shadowgraphy shaders, and novel oriented structure enhancement techniques to allow interactive visualization, exploration, and comparative analysis of scalar, vector, and time-varying volume datasets. Both traditional illustration techniques and photographic flow visualization techniques effectively reduce visual clutter by using compact oriented structure information to convey three-dimensional structures. Therefore, a key to the effectiveness of our system is using one-dimensional (Schlieren and shadowgraphy) and two-dimensional (silhouette) oriented structural information to reduce visual clutter, while still providing enough three-dimensional structural information for the users visual system to understand complex three-dimensional flow data. By combining these oriented feature visualization techniques with flexible transfer function controls, we can visualize scalar and vector data, allow comparative visualization of flow properties in a succinct, informative manner, and provide continuity for visualizing time-varying datasets.

Enhancing the Interactive Visualization of Procedurally Encoded Multifield Data with Ellipsoidal Basis Functions

Functional approximation of scattered data is a popular technique for compactly representing various types of datasets in computer graphics, including surface, volume, and vector datasets. Typically, sums of Gaussians or similar radial basis functions are used in the functional approximation and PC graphics hardware is used to quickly evaluate and render these datasets. Previously, researchers presented techniques for spatially‐limited spherical Gaussian radial basis function encoding and visualization of volumetric scalar, vector, and multifield datasets. While truncated radially symmetric basis functions are quick to evaluate and simple for encoding optimization, they are not the most appropriate choice for data that is not radially symmetric and are especially problematic for representing linear, planar, and many non‐spherical structures. Therefore, we have developed a volumetric approximation and visualization system using ellipsoidal Gaussian functions which provides greater compression, and visually more accurate encodings of volumetric scattered datasets. In this paper, we extend previous work to use ellipsoidal Gaussians as basis functions, create a rendering system to adapt these basis functions to graphics hardware rendering, and evaluate the encoding effectiveness and performance for both spherical Gaussians and ellipsoidal Gaussians.

DisplayCluster: An Interactive Visualization Environment for Tiled Displays

Display Cluster is an interactive visualization environment for cluster-driven tiled displays. It provides a dynamic, desktop-like windowing system with built-in media viewing capability that supports ultra high-resolution imagery and video content and streaming that allows arbitrary applications from remote sources (such as laptops or remote visualization machines) to be shown. This support extends to high-performance parallel visualization applications, enabling interactive streaming and display for hundred-mega pixel dynamic content. Display Cluster also supports multi-user, multi-modal interaction via devices such as joysticks, smart phones, and the Microsoft Kinect. Further, our environment provides a Python-based scripting interface to automate any set of interactions. In this paper, we describe the features and architecture of Display Cluster, compare it to existing tiled display environments, and present examples of how it can combine the capabilities of large-scale remote visualization clusters and high-resolution tiled display systems. In particular, we demonstrate that Display Cluster can stream and display up to 36 mega pixels in real time and as many as 144 mega pixels interactively, which is 3× faster and 4× larger than other available display environments. Further, we achieve over a gig pixel per second of aggregate bandwidth streaming between a remote visualization cluster and our tiled display system.

Hardware-assisted feature analysis and visualization of procedurally encoded multifield volumetric data

We take a new approach to interactive visualization and feature detection of large scalar, vector, and multifield computational fluid dynamics data sets that is also well suited for meshless CFD methods. Radial basis functions (RBFs) are used to procedurally encode both scattered and irregular gridded scalar data sets. The RBF encoding creates a complete, unified, functional representation of the scalar field throughout 3D space, independent of the underlying data topology, and eliminates the need for the original data grid during visualization. The capability of commodity PC graphics hardware to accelerate the reconstruction and rendering and to perform feature detection from this functional representation is a powerful tool for visualizing procedurally encoded volumes. Our RBF encoding and GPU-accelerated reconstruction, feature detection, and visualization tool provides a flexible system for visually exploring and analyzing large, structured, scattered, and unstructured scalar, vector, and multifield data sets at interactive rates on desktop PCs.

Time-Varying Data Visualization Using Functional Representations

In many scientific simulations, the temporal variation and analysis of features are important. Visualization and visual analysis of time series data is still a significant challenge because of the large volume of data. Irregular and scattered time series data sets are even more problematic to visualize interactively. Previous work proposed functional representation using basis functions as one solution for interactively visualizing scattered data by harnessing the power of modern PC graphics boards. In this paper, we use the functional representation approach for time-varying data sets and develop an efficient encoding technique utilizing temporal similarity between time steps. Our system utilizes a graduated approach of three methods with increasing time complexity based on the lack of similarity of the evolving data sets. Using this system, we are able to enhance the encoding performance for the time-varying data sets, reduce the data storage by saving only changed or additional basis functions over time, and interactively visualize the time-varying encoding results. Moreover, we present efficient rendering of the functional representations using binary space partitioning tree textures to increase the rendering performance.

Abstracting Attribute Space for Transfer Function Exploration and Design

Currently, user centered transfer function design begins with the user interacting with a one or two-dimensional histogram of the volumetric attribute space. The attribute space is visualized as a function of the number of voxels, allowing the user to explore the data in terms of the attribute size/magnitude. However, such visualizations provide the user with no information on the relationship between various attribute spaces (e.g., density, temperature, pressure, x, y, z) within the multivariate data. In this work, we propose a modification to the attribute space visualization in which the user is no longer presented with the magnitude of the attribute; instead, the user is presented with an information metric detailing the relationship between attributes of the multivariate volumetric data. In this way, the user can guide their exploration based on the relationship between the attribute magnitude and user selected attribute information as opposed to being constrained by only visualizing the magnitude of the attribute. We refer to this modification to the traditional histogram widget as an abstract attribute space representation. Our system utilizes common one and two-dimensional histogram widgets where the bins of the abstract attribute space now correspond to an attribute relationship in terms of the mean, standard deviation, entropy, or skewness. In this manner, we exploit the relationships and correlations present in the underlying data with respect to the dimension(s) under examination. These relationships are often times key to insight and allow us to guide attribute discovery as opposed to automatic extraction schemes which try to calculate and extract distinct attributes a priori. In this way, our system aids in the knowledge discovery of the interaction of properties within volumetric data.

Data-parallel mesh connected components labeling and analysis

We present a data-parallel algorithm for identifying and labeling the connected sub-meshes within a domaindecomposed 3D mesh. The identification task is challenging in a distributed-memory parallel setting because connectivity is transitive and the cells composing each sub-mesh may span many or all processors. Our algorithm employs a multi-stage application of the Union-find algorithm and a spatial partitioning scheme to efficiently merge information across processors and produce a global labeling of connected sub-meshes. Marking each vertex with its corresponding sub-mesh label allows us to isolate mesh features based on topology, enabling new analysis capabilities. We briefly discuss two specific applications of the algorithm and present results from a weak scaling study. We demonstrate the algorithm at concurrency levels up to 2197 cores and analyze meshes containing up to 68 billion cells.

Ray tracing and volume rendering large molecular data on multi-core and many-core architectures

Visualizing large molecular data requires efficient means of rendering millions of data elements that combine glyphs, geometry and volumetric techniques. The geometric and volumetric loads challenge traditional rasterization-based vis methods. Ray casting presents a scalable and memory- efficient alternative, but modern techniques typically rely on GPU-based acceleration to achieve interactive rendering rates. In this paper, we present bnsView, a molecular visualization ray tracing framework that delivers fast volume rendering and ball-and-stick ray casting on both multi-core CPUs and many-core Intel® Xeon Phi™ co-processors, implemented in a SPMD language that generates efficient SIMD vector code for multiple platforms without source modification. We show that our approach running on co- processors is competitive with similar techniques running on GPU accelerators, and we demonstrate large-scale parallel remote visualization from TACCs Stampede supercomputer to large-format display walls using this system.

Remote visualization of large scale data for ultra-high resolution display environments

ParaView is one of the most widely used scientific tools that support parallel visualization of large scale data. The Scalable Adaptive Graphics Environment (SAGE) is a graphics middleware that enables real-time streaming of ultra-high resolution visual content from distributed visualization resources to scalable tiled displays connected by ultra-high-speed networks. Integrating these two technologies enables visualization of large-scale data at an extremely high resolution to be displayed on distantly located scalable tiled displays. The benefits, limitations, and future directions for this approach will be discussed.

RBF Volume Ray Casting on Multicore and Manycore CPUs

Modern supercomputers enable increasingly large N‐body simulations using unstructured point data. The structures implied by these points can be reconstructed implicitly. Direct volume rendering of radial basis function (RBF) kernels in domain‐space offers flexible classification and robust feature reconstruction, but achieving performant RBF volume rendering remains a challenge for existing methods on both CPUs and accelerators. In this paper, we present a fast CPU method for direct volume rendering of particle data with RBF kernels. We propose a novel two‐pass algorithm: first sampling the RBF field using coherent bounding hierarchy traversal, then subsequently integrating samples along ray segments. Our approach performs interactively for a range of data sets from molecular dynamics and astrophysics up to 82 million particles. It does not rely on level of detail or subsampling, and offers better reconstruction quality than structured volume rendering of the same data, exhibiting comparable performance and requiring no additional preprocessing or memory footprint other than the BVH. Lastly, our technique enables multi‐field, multi‐material classification of particle data, providing better insight and analysis.