Volume rendering is a technique for visualizing 3D arrays of sampled data. It has applications in areas such as medical imaging and scientific visualization, but its use has been limited by its high computational expense. Early implementations of volume rendering used brute-force techniques that require on the order of 100 seconds to render typical data sets on a workstation. Algorithms with optimizations that exploit coherence in the data have reduced rendering times to the range of ten seconds but are still not fast enough for interactive visualization applications. In this thesis we present a family of volume rendering algorithms that reduces rendering times to one second. First we present a scanline-order volume rendering algorithm that exploits coherence in both the volume data and the image. We show that scanline-order algorithms are fundamentally more efficient than commonly-used ray casting algorithms because the latter must perform analytic geometry calculations (e.g. intersecting rays with axis-aligned boxes). The new scanline-order algorithm simply streams through the volume and the image in storage order. We describe variants of the algorithm for both parallel and perspective projections and

This paper presents a divide-and-conquer ray-traced volume rendering algorithm and a parallel image compositing method, along with their implementation and performance on the Connection Machine CM-5, and networked workstations. This algorithm distributes both the data and the computations to individual processing units to achieve fast, high-quality rendering of high-resolution data. The volume data, once distributed, is left in place, regardless of viewing position. The processing nodes perform local raytracing of their subvolume concurrently. No communication between processing units is needed during this local ray-tracing process. A subimage is generated by each processing unit and the final image is obtained by compositing subimages in the proper order, which can be determined a priori. Composition is done in parallel via a new algorithm we call BinarySwap compositing. Test results on both the CM-5 and a group of networked workstations demonstrate the practicality of our rendering ...

In computer graphics, rendering is the process by which an abstract description of a scene is converted to an image. When the scene is complex, or when high-quality images or high frame rates are required, the rendering process becomes computationally demanding. To provide the necessary levels of performance, parallel computing techniques must be brought to bear. Although parallelism has been exploited in computer graphics since the early days of the field, its initial use was primarily in specialized applications. The VLSI revolution of the late 1970Õs and the advent of scalable parallel computers during the late 1980Õs changed this situation. Today, parallel hardware is routinely used in graphics workstations, and numerous software-based rendering systems have been developed for general-purpose parallel architectures. This article provides a broad introduction to the subject of parallel rendering, encompassing both hardware and software systems. The focus is on the underlying concepts and the issues which arise in the design of parallel rendering algorithms and systems. We examine the different types of parallelism and how they can be applied in rendering applications. Concepts from parallel computing, such as data decomposition, task granularity, scalability, and load balancing, are considered in relation to the rendering

visualization, but also for modeling objects and structures derived from volumetric data. The paper describes work in progress towards demonstrating the utility of a voxel-based format for modeling physical interactions between virtual objects. Data structures are presented that help to optimize storage requirements and preserve object integrity during object movement. An adaptation to volume rendering algorithms is discussed that enables objects to be rendered individually and then combined in a final compositing step. Finally, algorithms and prototype systems are presented that use a voxel-based format to model physical interactions between objects. These physical interactions include collision detection and avoidance of object interpenetration, haptic, or tactile, exploration of virtual objects using a force feedback device, and object deformation.

This paper examines the many ways to structure parallel volume rendering algorithms and analyzes the communication costs associated with them. Parallel volume rendering algorithms are enumerated through a taxonomy which sorts them into two main classes that exhibit similar communication costs: image and object partitions. The intrinsic communication costs for algorithms in these classes are analyzed independent of an implementation. Given a network model for a target system, an algorithm's intrinsic communication cost can be used to estimate the time consumed by communication and the effect upon communication time as the system size and data size are varied. Communication cost and time are measured on the Intel Touchstone Delta to verify the predicted scaling behavior. The results show that, for a fixed screen size, systems with mesh networks scale well for object partition algorithms - the time required for communication decreases as the data and system sizes increase. 1. Introductio...

This paper describes work-in-progress on developing parallel visualization strategies for 3D Adaptive Mesh Refinement (AMR) data. AMR is a simple and powerful tool for modeling many important scientific and engineering problems. However, visualization tools for 3D AMR data are not generally available. Converting AMR data onto a uniform mesh would result in high storage requirements, and rendering the uniform-mesh data on an average graphics workstation can be painfully slow if not impossible. The adaptive nature of the embedded mesh demands sophisticated visualization calculations. In this work, we compare the performance and storage requirements of a parallel volume renderer for regular-mesh data with a new parallel renderer based on adaptive sampling. While both renderers can achieve interactive visualization, the new approach offers significant performance gains, as indicated by our experiments on the SGI/Cray T3E. 1

We describeatechnique for achieving fast volume ray casting on parallel machines, using a load balancing scheme and an e#cient pipelined approach to compositing. We propose a new model for measuring the amount of work one needs to perform in order to render a given volume, and use this model to obtain a better load balancing scheme for distributed memory machines. We also discuss in detail the design tradeo #s of our technique. In order to validate our model we have implemented it on the Intel iPSC#860 and the Intel Paragon, and conducted a detailedperformance analysis. 1 Introduction As researchers and engineers use volume rendering to study complex physical and abstract structures they need a coherent, powerful, easy to use visualization tool, that lets them interactively change all the necessary parameters. Unfortunately,even with the latest volume rendering acceleration techniques running on top-of-the-line workstations, it still takes a few seconds to a few minutes to volume ren...

Figure 1: Views of the head section (512x512x209) of the visible female CT data with 16 nodes (a space has been left between the subvolumes to highlight their boundaries). Using a 3 years old 32-node COTS cluster, a volume dataset can be rendered at constant 13 frames per second on a 1024 × 768 rendering area using 5 nodes. On a 1.5 years old, fully optimized, 5-node COTS cluster, the frame rate obtained for the same rendering area reaches constant 31 frames per second. We truly expect our future work, including further algorithm optimizations and hardware tuning on a modern PC cluster, to provide higher frame rates for bigger datasets (using more nodes) on larger rendering areas. Sort-last parallel rendering is an efficient technique to visualize huge datasets on COTS clusters. The dataset is subdivided and distributed across the cluster nodes. For every frame, each node renders a full resolution image of its data using its local GPU, and the images are composited together using a parallel image compositing algorithm. In this paper, we present a performance evaluation of standard sort-last parallel rendering methods and of the different improvements proposed in the literature. This evaluation is based on a detailed analysis of the different hardware and software components.

This paper presents a parallel volume rendering system, ParVox, for large volumes of 4-D data sets in regular structured grids. A parallel volume rendering API based on the splatting algorithm constitutes the core of the ParVox system. A network interface program takes commands from an X Window based GUI, calls the API to perform the rendering functions, compresses the rendered images and sends them back to the GUI window. The ParVox system is designed for interactive, distributed visualization of large multiple time steps, multiple parameters volume datasets. The parallel splatting algorithm employs both object space decomposition and image space decomposition; an asynchronous image compositing scheme based on the direct send model reduces both the communication overhead and the synchronization overhead. The ParVox system architecture, the parallel algorithm and its implementation on the Cray T3D and the parallel wavelet compression algorithm are discussed extensively in this paper. The performance results and some optimization techniques are also presented.

We describe the design and implementation of volume rendering algorithms in a distributed and collaborative software environment. The algorithms use the computational power of a heterogeneous cluster of workstations on a network to produce translucent or shaded images of extremely large volume data sets. Graphics functions are performed using a machine independent 3D graphics and windows library, running on multiple platforms in a heterogeneous environment. We also describe a synchronously conferenced application which wehavebuilt to support collaborative visualization, allowing multiple users to share and interact over a volume data set while viewing multiple renderings with independent viewing directions, cutaways, shading parameters, etc.