Jane Wilhelms, Allen Van Gelder, Paul Tarantino, and Jonathan Gibbs. Hierarchical and parallelizable direct volume rendering for irregular and multiple grids. In Visualization, pages 57--65, 1996.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....determines if the face is intersected by the ray. So his scheme tests two faces for each tetrahedral cell to determine the exit point on the average whereas Garrity s scheme [8] always tests three faces. Bunyk et al. 1] present a simple and efficient ray casting algorithm that uses ideas from [8, 28, 10]. The algorithm essentially breaks the cells into their corresponding faces and visibility determination is performed after the faces have been transformed into screen space. The actual ray casting is performed independently for each pixel by performing a walk in the cell complex that is stored as ....

....into an x bucket with respect to their x coordinates in increasing order. When processing pixels in the current scanline, an active face list is created for the current pixel using the x bucket. In this way, the number of faces to be tested for intersection is reduced considerably. Wilhelms et al. [28] propose a similar algorithm for hierarchical and parallelizable rendering of unstructured grids. Giertsen [9] utilizes a 2D scan plane buffer to store information within the plane perpendicular to a scanline. In this approach, z dimension is discretized due to the scan plane buffer. The ....

J. Wilhelms, A. V. Gelder, P. Tarantino, and J. Gibbs, "Hierarchical and parallelizable direct volume rendering for irregular and multiple grids," in Proceedings of the IEEE Visualization'96, (, ed.), (), pp. 57--64, 1996.

....parts of the algorithm dependent on the type of grid and the corresponding data structures will be described in sections IV and V. A. Principle Incremental slicing, first introduced for volume rendering by Yagel [30] reduces the 3D volume rendering problem to a set of 2D problems, as in [46] [47], 35] 17] More precisely, the volume is sliced by a sweeping plane parallel to the screen in back to front order. The intersection of the cells with a slicing plane results in a set of polygons, which can be rendered in alpha blending mode with graphics hardware. Thus, the algorithm can be ....

J. P. Wilhelms, A. V. Gelder, P. Tarantino, and J. Gibbs, "Hierarchical and parallelizable direct volume rendering for irregular and multiple grids," IEEE Visualization '96, pp. 57--64, October 1996.

.... of the last years have produced several efficient techniques [29, 24, 25, 5, 21] see [19] for a more complete list) At the same time researchers have proposed strategies for efficient visualization of unstructured volume meshes using screen based ray casting [8, 3, 34] or objectbased sweeping [32, 6, 18] (see [7] for a survey about rendering unstructured volume grids) The basic ingredients of unstructured hexahedral volume meshes can be classified into three things: mesh connectivity, that is the incidence relation among the vertices, edges, faces, and hexahedra, mesh geometry, that is the 3D ....

J. Wilhelms, A. V. Gelder, P. Tarantino, and J. Gibbs. Hierarchical and parallelizable direct volume rendering for irregular and multiple grids. In Visualization'96 Conf. Proc., pages 57--64, 1996.

.... of the last years have produced several efficient techniques [29, 24, 25, 5, 20] see [21] for a more complete list) At the same time researchers have proposed strategies for efficiently visualization of unstructured volume meshes using screen based ray casting [8, 3, 34] or object based sweeping [32, 6, 18] (see [7] for a survey about rendering unstructured volume grids) The basic ingredients of unstructured hexahedral volume meshes can be classified into three things: mesh connectivity, that is the incidence relation among the vertices, edges, faces, and hexahedra, the mesh geometry, that is the ....

J. Wilhelms, A. V. Gelder, P. Tarantino, and J. Gibbs. Hierarchical and parallelizable direct volume rendering for irregular and multiple grids. In Visualization'96 Conference Proceedings, pages 57--64, 1996.

....random access to the cells, connectivity information, and in some cases extra memory (such as [5] to optimize the computation of intersections of rays with faces of the cell complex. Other techniques use scan line algorithms, which sweep the data with a plane perpendicular to the image plane [58, 6]. Some of them (e.g. 48] are designed to be memory efficient, but still use the connectivity of the mesh. Others (e.g. 21, 55] use discrete buffers in to determine the order of compositing, and completely avoid the need for connectivity information. However, using discrete buffers has the ....

....very efficient ray casting based rendering algorithm for curvilinear grids, and parallelized their ray caster using an image based task scheduling scheme. The parallelization of a ray casting technique has also been studied by Uselton [52] with very good results. Challinger [6] and Wilhelms et al. [58] proposed scan line rendering algorithms. Both papers report on parallelization, which is the main focus of [6] Challinger [6] also used an image tiling scheme for parallelization with very good results. Other parallel techniques on shared memory machines include the results by Williams [59] ....

J. P. Wilhelms, A. Van Gelder, P. Tarantino, and J. Gibbs. Hierarchical and parallelizable direct volume rendering for irregular and multiple grids. IEEE Visualization '96, pages 57--64, October 1996.

....space, which allows to avoid complex ray intersection computations. The major drawback of screen space methods is the fact that the computation time strongly depends on the image size. In object space methods, the cells of the grid are sorted (in depth order) then projected on the screen [14] [26], 4] Several approaches have been proposed to project the cells on the screen. In splatting approaches, a filter is applied to the cells when they are projected to the screen ( 13] 24] In [27] an accurate method is proposed for meshes composed of linear or quadratic tetrahedra. In [23] a ....

Jane P. Wilhelms, Allen Van Gelder, Paul Tarantino, and Jonathan Gibbs. Hierarchical and parallelizable direct volume rendering for irregular and multiple grids. IEEE Visualization '96, pages 57--64, October 1996. ISBN 0-89791-864-9.

.... 62] 31, 75, 30, 38, 86] 41, 21, 119, 112, 118, 45, 58] 55, 65, 14, 29, 79] 115, 53, 70, 74, 87, 19, 117, 32] 72, 57, 92, 36, 7] 111, 12, 90, 97, 78, 5] 15, 27, 35, 43, 87, 28, 51, 22, 23, 59, 60, 64, 63, 66, 68, 71, 80, 81, 91, 93, 99, 47, 108] 103, 116, 83, 106, 107, 105, 2, 76, 110, 77] [11, 104, 101, 96, 89, 26, 73, 6, 17, 52, 54, 85, 114, 9, 82, 37, 24] [8, 61, 3, 56, 67, 88, 44, 10, 69, 48, 16, 84, 113, 95, 13, 50, 94, 40, 46] 100, 109, 102] 25, 18, 33, 39] ....

Jane Wilhelms, Allen Van Gelder, Paul Tarantino, and Jonathan Gibbs. Hierarchical and parallelizable direct volume rendering for irregular and multiple grids. In Proceedings of Visualization, pages 57--64, San Francisco, CA, October 1996. 7

....according to the actual viewing definition and the determination of intersection points between rays and cells is accelerated by this approach. No optimization of the integration process itself will be achieved. Our method is very similar to the ones proposed by Giertsen [4] Silva [14] Wilhelms [22], and Yagel [24] All three approaches utilize the coherence within cutting planes in object space and thus reduce the 3D problem to a 2D problem within each plane. Furthermore, the sorting procedure within each plane can be reduced to the sorting of primitives along each line of sight. However, ....

....and sorted with respect to the actual view y direction. From the sorted list those cells can be obtained which contribute to the actual scanline by successively updating a y active list which holds all active vertices for the scanline. A very complete description of this procedure can be found in [4, 14, 22]. The second approach, as proposed in [22] uses a k d tree that is build over all primitives. At every node of the tree the bounding box structure for all primitives inside the node is stored. Polygons are inserted at the node deepest in the tree that completely contains the polygon. Each time ....

[Article contains additional citation context not shown here]

J. Wilhelms, A. van Gelder, P. Tarantino, and J. Gibbs. Hierarchical and parallelizable direct volume rendering for irregular and multiple grids. In Visualization 1996, pages 57--65. IEEE, 1996.

....a state of the art scientific visualization technique closer to the computational scientist. 2 Rendering Algorithm Here, we describe our technique as it applies to tetrahedral grids. The extension to other cell types is fairly simple. The basic idea is very simple, and similar to the work of [17, 2, 4]. Instead of rendering cells, we render their faces (or triangulation of their faces, in the case of general cell data) We use ray casting for our rendering. The actual depth sorting is based on the implicit ordering provided by the connectivity information in the cells (similar to the method ....

J. Wilhelms, A. Van Gelder, P. Tarantino, and J. Gibbs. "Hierarchical and Parallelizable Direct Volume Rendering for Irregular and Multiple Grids," IEEE Visualization '96, pp. 57--64, 1996.

....a state of the art scientific visualization technique closer to the computational scientist. 2 Rendering Algorithm Here, we describe our technique as it applies to tetrahedral grids. The extension to other cell types is fairly simple. The basic idea is very simple, and similar to the work of [17, 2, 4]. Instead of rendering cells, we render their faces (or triangulation of their faces, in the case of general cell data) We use ray casting for our rendering. The actual depth sorting is based on the implicit ordering provided by the connectivity information in the cells (similar to the method ....

J. Wilhelms, A. Van Gelder, P. Tarantino, and J. Gibbs. "Hierarchical and Parallelizable Direct Volume Rendering for Irregular and Multiple Grids," IEEE Visualization '96, pp. 57--64, 1996.

....a more elaborate approach for rendering general irregular grids. Evolving from Giertsen s sweep plane algorithm, their new strategy is careful to exploit spatial coherence. The algorithm is potentially parallelizable but tests were only performed on a Sun UltraSPARC 1. In addition, Wilhelms et al. [22] developed a hierarchical and parallelizable volume rendering technique for irregular and multiple grids. This algorithm favors coarse grain parallelism for a shared memory MIMD architecture. Palmer and Taylor [16] devised a true distributed memory ray casting volume renderer for unstructured ....

....regions can also serve as workload units should we ever need to perform dynamic load balancing. 5 Rendering Direct volume rendering algorithms can be classified into either ray casting [6] 11] 12] 20] or projection methods [15] 17] Projection methods may be further categorized as cellprojected [22], slice projected [24] or vertex projected [14] We have chosen a cell projection method similar to [15] because it offers more flexibility in data distribution and is more accurate. 8 During rendering, processors follow the same path through the spatial partitioning tree, processing all of the ....

Jane Wilhelms, Allen Van Gelder, Paul Tarantino, and Jonathan Gibbs. Hierarchical and parallelizable direct volume rendering, Proceedings Visualization `96, IEEE CS Press, October 1996, pp. 57--64.

....data resolution. A fast and adaptive visualization of volume data is implemented in the hierarchical splatting algorithm by Laur and Hanrahan [37] They have used a L 2 type error indicator on an octree encoded voxel data base to speed up rendering substantially. Wilhelms and van Gelder [60] use such a hierarchical speed up in a scan plane type approach to volume rendering, especially for preview purposes. Cignoni et al. 8] have applied a successive adaptive refinement of volumes by Delaunay methods, which leads to non nested hierarchical meshes. They discuss further issues in [9] ....

J. P. Wilhelms, A. Van Gelder, P. Tarantino, and J. Gibbs. Hierarchical and parallelizable direct volume rendering for irregular and multiple grids. In Proceedings Visualization, 1996.

....Figure 1.1 displays a surface mesh used in the high lift analysis work [8] and Figure 1.2 shows a close up view of the same mesh. The corresponding volume mesh would be too cluttered to view directly. Visualizing unstructured grid data has been an active area of research in recent years [2, 5, 6, 7, 9, 10, 11, 12, 13]. However, interactive performance for high fidelity visualization of large datasets, such as the high lift analysis solutions, can only be obtained with the help of parallel computers. At ICASE, we have been developing a parallel volume renderer for unstructured grid data. Our design is based on ....

J. WILHELMS,A.VAN GELDER,P.TARANTINO, AND J. GIBBS (1996), Hierarchical and parallelizable direct volume rendering for irregular and multiple grids, in Proceedings Visualization `96, IEEE CS Press, pp. 57--64.

....works in two phases: a space sweep, with a sweep plane orthogonal to the viewing (XY) plane, and a second sweep on that plane with a sweep line parallel to the Z axis. Another scanline algorithm that exploits a spatial hierarchical organization of the dataset was described by Wilhelms et al. [39]. The main difference of this approach (which like the previous algorithms uses both a scan plane orthogonal to the view plane and a scan line lying on that plane) is that it renders semi transparent regions of space between cell faces as well as opaque polygonal surfaces immersed in the dataset. ....

J. Wilhelms, A. Van Gelder, P. Tarantino, and J. Gibbs. Hierarchical and parallelizable direct volume rendering for irregular and multiple grids. In R. Yagel G. Nielson, editor, Visualization '96 Proceedings, pages 57--64. IEEE Press, 1996.

....splatting, cell projection, visibility ordering, depth sorting. 1 Introduction Typically unstructured meshes have a complex geometric configuration and the mathematics of the absorption emission integral are quite complex. Therefore most existing volume rendering systems for unstructured data [5, 6, 7, 8, 10, 12, 13, 15, 16, 23, 29, 30, 31, 33, 34, 35, 37, 41] introduce various simplifying assumptions and approximations into the algorithm in order to cope with these complexities in an efficient manner. Another aspect of unstructured meshes is that typically they are adaptively refined, so that in areas where the field is changing rapidly, the cells ....

J. Wilhelms, A. Van Gelder, P. Tarantino and J. Gibbs, "Hierarchical and Parallelizable Direct Volume Rendering for Irregular and Multiple Grids," Proc. Visualization '96, pp. 57--64, Oct. 1996.

....are presented where no depth ordering is strictly necessary, and in some cases calculated approximately. Very fast rendering is achieved by using graphics hardware to project the partially sorted faces. A recent scanline technique that handles multiple, and overlapping grids is presented in [44]. They process the set of polygonal facets of cells, by first bucketing them according to which scanline contains the topmost vertex, and then maintaining a y actives list of polygons present at each scanline, as they sweep from top to bottom (in y) Then, on each scanline, they scan in x, ....

J. Wilhelms, A. Van Gelder, P. Tarantino, and J. Gibbs. "Hierarchical and Parallelizable Direct Volume Rendering for Irregular and Multiple Grids," IEEE Visualization '96, pp. 57--64, 1996.

....80 and 95 vertex decimation to the original volume. It takes under two minutes to decimate the original volume by these amounts. 4.1. Effect of Decimation on Direct Volume Rendering We have used a direct volume rendering system based on a generalized software scan conversion of polygons [11]. The scan conversion algorithm generalizes traditional polygon scan line methods in that it renders semi transparent regions of space between polygons, as well as opaque surfaces. This method requires no graphics hardware, and produces excellent quality images. We have found that a volume ....

....rendering time does not decrease significantly. For a volume decimated using the massmetric by 80 , the rendering time decreases by about 27 . A volume decimated by 95 can generate images with few artifacts and the rendering time decreases by about 55 . This is due to our direct volume renderer [11], for which rendering time is proportional to the area of polygons rendered. While our decimation greatly reduces the number of geometrical objects, it does not reduce the area of the retained triangles proportionately. The plots in Figure 7 show that the area of the retained triangles does not ....

J. Wilhelms, A. Van Gelder, P. Tarantino, and J. Gibbs. Hierarchical and parallelizable direct volume rendering for irregular and multiple grids. In Proceedings of Visualization '96, pages 57--64, October 1996.

No context found.

Jane Wilhelms, Allen Van Gelder, Paul Tarantino, and Jonathan Gibbs. Hierarchical and parallelizable direct volume rendering for irregular and multiple grids. In Visualization, pages 57--65, 1996.

No context found.

Wilhelms, J., Gelder, A., Tarantino, P., Gibbs, J.: Hierarchical and parallelizable direct volume rendering of irregular and multiple grids. In Proceedings IEEE Visualization 1996 pp. 57--65

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC