Datasets:
id int64 0 2.79k | image imagewidth (px) 220 2.25k | caption stringlengths 41 1.6k | referencing_paragraphs listlengths 1 10 | figure_number stringclasses 24
values | title stringlengths 8 137 | year int64 2k 2.02k | doi stringlengths 21 27 | paper_url stringlengths 39 61 | conference stringclasses 2
values | vis_type int64 15 100 | encoding_type stringclasses 72
values | dim_type stringclasses 6
values | vis_url stringlengths 56 63 | cap_url stringlengths 58 65 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0 | Figure 7: (a) Human (256^3); (b) Macaque (256^3); (c) Erectus (128^3); (d) Human-Macaque (128^3); (e) Human-Macaque (256^3); (f) Erectus-Human (128^3); (g) Human-Chimpanzee (256^3); (h) Human-Macaque-Human (256^3); (i) Erectus-Macaque (128^3). | [
"Finally, if the deformation function is the identity, i. e. no deformation is intended, the target blocks would be the same as the texture blocks. Using the above template based approach we can directly project the texture blocks without worrying about the texture association problem and the repeated texture bindi... | 7 | Deformable volume rendering by 3D texture mapping and octree encoding | 1,996 | 10.1109/VISUAL.1996.567609 | http://dx.doi.org/10.1109/VISUAL.1996.567609 | Vis | 100 | null | null | |||
1 | Figure 9 : Surface View of Ozone from 3 sites in London. (Provisional Data) | [
"Some example output is shown here where the user has selected from the forms interface to view Ozone data from London covering the 25 day period from June 1st 0 hours to June 25th 23 hours using a 1 day period on the x-axis. Two views are possible: a block chart (Figure 8), with pillars indicating the discrete dat... | 9 | Visualization over the World Wide Web and its application to environmental data | 1,996 | 10.1109/VISUAL.1996.567610 | http://dx.doi.org/10.1109/VISUAL.1996.567610 | Vis | 100 | null | null | |||
2 | Figure 3: Two-dimensional texture map used to implement Phongβs reflection model for line segments. Parameter values are k_a = 0.1, k_d = 0.3, k_s = 0.6, and n = 40. | [
"In order to obtain the correct light intensity corresponding to L Β· T = 2 t_1 - 1 and V Β· T = 2 t_2 - 1 we can use a two-dimensional texture map P(t_1, t_2) . Adding a constant ambient term k_a as well as the diffuse contribution from Eq. (4) we can perform the whole shading calculation for a single light source i... | 3 | Interactive visualization of 3D-vector fields using illuminated stream lines | 1,996 | 10.1109/VISUAL.1996.567777 | http://dx.doi.org/10.1109/VISUAL.1996.567777 | Vis | 15 | null | null | |||
3 | Fig 4a: Single frequency noise image for Pacific Ocean flow (left). The LIC image (right) uses a filter kernel length of 40 pixels both forward and backward to produce thin streaks along streamlines. | [
"How well does this strategy of mapping large features to large velocities work on empirical data from real vector fields? The fourth example shows multi-frequency noise derived from a more complex vector field: the Pacific Ocean's surface velocity (Figure 4). In figure 4 we used 8 noise levels and a convolution ke... | 4 | Multi-frequency noise for LIC | 1,996 | 10.1109/VISUAL.1996.567784 | http://dx.doi.org/10.1109/VISUAL.1996.567784 | Vis | 15 | null | null | |||
4 | Fig 4b: Multi-frequency noise image for Pacific Ocean flow (left) using 8 noise levels. The LIC image (right) uses a filter length that ranges from 5 pixels (for the smallest velocities) to 40 pixels (for the largest velocities) both forward to produce variable-resolution streaks along streamlines. | [
"How well does this strategy of mapping large features to large velocities work on empirical data from real vector fields? The fourth example shows multi-frequency noise derived from a more complex vector field: the Pacific Ocean's surface velocity (Figure 4). In figure 4 we used 8 noise levels and a convolution ke... | 4 | Multi-frequency noise for LIC | 1,996 | 10.1109/VISUAL.1996.567784 | http://dx.doi.org/10.1109/VISUAL.1996.567784 | Vis | 15 | null | null | |||
5 | Figure 4: Simulation I: 4/100 time steps. All features are tracked and color coded. (See color images) | [
"Octree features can be visualized using volume rendering or fitting a surface around the boundary. An example is shown in Figure 3 and 4. The upper left image of Figure 3 shows a 128Β³ turbulent dataset visualized using standard isosurfaces. In Figure 4, the same dataset is shown. However, segmentation was first pe... | 4 | Volume tracking | 1,996 | 10.1109/VISUAL.1996.567807 | http://dx.doi.org/10.1109/VISUAL.1996.567807 | Vis | 15 | null | null | |||
6 | Figure 5: Feature tracking (Simulation I). The evolution of the light green feature (from t=1) is highlighted. All other features are colored grey. When an object is about to merge with the light green one it is given a different color (as in t=4). (See color images.) | [
"Since all regions are tracked and all events are classified, any of one of these can be mapped back to the visualization. In Figure 5, the light green feature of Figure 4 is tracked. Here, 16/100 datasets are shown with one feature colored and all others in grey. When a feature is about to merge with the green one... | 5 | Volume tracking | 1,996 | 10.1109/VISUAL.1996.567807 | http://dx.doi.org/10.1109/VISUAL.1996.567807 | Vis | 15 | null | null | |||
7 | Figure 1: The problem of reconstructing a body from cross-sections can also be viewed as a contour blending problem, where the third dimension is time. | [
"The problem of 2D shape blending can also be considered as a problem of body reconstruction from cross-sections [18], as demonstrated in Figure 1. Some reconstruction methods use bivariate interpolation after finding a correspondence between the contours of the cross-sections. When the cross-sections are not dense... | 1 | Contour blending using warp-guided distance field interpolation | 1,996 | 10.1109/VISUAL.1996.567812 | http://dx.doi.org/10.1109/VISUAL.1996.567812 | Vis | 15 | null | null | |||
8 | Figure 5: Warp-Guided DFI. (a) The source Distance Field warped towards the target. (c) The target Distance Field warped towards the source. (b) The intermediate interpolated Distance Field. The contours are highlighted in white and the distances appear in absolute values. | [
"Figures 6, 7 and 8 show three sequences of blending between 2D contours. The algorithm is given either a set of two images of the two contours, as in Figures 6 and 7, or as in Figure 8, where the black pixels are interpreted as the object, and the contour is the boundary of the object. We then apply a 2D distance ... | 5 | Contour blending using warp-guided distance field interpolation | 1,996 | 10.1109/VISUAL.1996.567812 | http://dx.doi.org/10.1109/VISUAL.1996.567812 | Vis | 15 | null | null | |||
9 | Figure 4: The complex dendritic paths of an LGN cell are explored through visual and force feedback. | [
"Figure 4 illustrates the use of the system in understanding a complex set of dendrites emanating from a lateral geniculate nucleus (LGN) cell. This voxel LGN cell was scanned with a confocal microscope, and volume and haptic rendered as discussed in Section 4.1. The ability to feel as well as see the dendrites pro... | 4 | A haptic interaction method for volume visualization | 1,996 | 10.1109/VISUAL.1996.568108 | http://dx.doi.org/10.1109/VISUAL.1996.568108 | Vis | 15 | null | null | |||
10 | Figure 8: Haptic modeling tools are used to construct a volume rendered tree and clouds. | [
"Figure 8 shows a volumetric scene created from an empty volume with a resolution of voxels. Several construction and painting tools were used, and the image was generated using a volume rendering technique. This haptic interaction method can form the basis of a volume modeling [17,5] or painting [1,7] system that ... | 8 | A haptic interaction method for volume visualization | 1,996 | 10.1109/VISUAL.1996.568108 | http://dx.doi.org/10.1109/VISUAL.1996.568108 | Vis | 15 | null | null | |||
11 | Figure 2. a) Initial tessellation of solvent-accessible surface and b) result of applying triangular texture to that surface. | [
"Obviously, the most regular texture patterns will be formed over surfaces with uniformly sized triangles. On such surfaces, texture elements are replicated at a constant size across the surface. Additionally, since the texture element is defined on an equilateral triangle, the textured surface will appear most reg... | 2 | Opacity-modulating triangular textures for irregular surfaces | 1,996 | 10.1109/VISUAL.1996.568111 | http://dx.doi.org/10.1109/VISUAL.1996.568111 | Vis | 15 | null | null | |||
12 | Figure 3. a) Regularized tessellation of solvent-accessible surface and b) result of applying triangular texture to the new surface. | [
"[Turk92]. First, a new set of points is chosen at random locations on the surface. These points will later become the vertices of the regularized tessellation. Then, a relaxation procedure is applied to move points away from neighboring points by way of a simulated repulsion force. The result of this process is a ... | 3 | Opacity-modulating triangular textures for irregular surfaces | 1,996 | 10.1109/VISUAL.1996.568111 | http://dx.doi.org/10.1109/VISUAL.1996.568111 | Vis | 100 | null | null | |||
13 | Figure 4: Shift node perturbation: a. nodes are chosen to be removed in a way that will not change the knot type, b. nodes are placed elsewhere on the knot, c. the knot is evolved according to he equations of motion. | [
"The second perturbation method is the node shift method which tightens and loosens different parts of the knot. Nodes, or vertices, are shifted along the polygonal knot followed by the application of spring force and edge repulsion forces to the knot. With the newly shifted positions of nodes and the evolution of ... | 4 | Untangling Knots by Stochastic Energy Optimization | 1,996 | 10.1109/VISUAL.1996.568120 | http://doi.ieeecomputersociety.org/10.1109/VISUAL.1996.568120 | Vis | 100 | null | null | |||
14 | Figure 7: Distribution of the energy gradient magnitude (i.e., strength of the self-repulsive force) along the curve for the initial (top) and an intermediate (bottom) configuration in the iterative improvement approach. The right column shows the configuration on the left after it was unfolded into its Gauss model. Re... | [
"Case 2 is an unknot represented by 139 vertices whose tangled configuration was described by Ochiai [12] (see top row of the color plate for a stereo pair). The energy of the initial configuration was 4464.47. Gradient descent technique was used to untangle this configuration. Care had to be taken in selecting a s... | 7 | Untangling Knots by Stochastic Energy Optimization | 1,996 | 10.1109/VISUAL.1996.568120 | http://doi.ieeecomputersociety.org/10.1109/VISUAL.1996.568120 | Vis | 100 | null | null | |||
15 | Figure 10: Simplification of an object consisting of 1341 vertices and 2449 triangles. The size of its bounding box is 20 Γ20 Γ20 units. The approximation error is 0.25 units. The simplified mesh contains 124 vertices and 208 triangles. This is a reduction rate of 91%. Note the preservation of the sharp edges. | [
"This strategy of implicit sorting preserves sharp edges in the original triangulation, see Figure 10."
] | 10 | Mesh reduction with error control | 1,996 | 10.1109/VISUAL.1996.568124 | http://dx.doi.org/10.1109/VISUAL.1996.568124 | Vis | 15 | null | null | |||
16 | Figure 1: Use of clipping planes to improve rendering speed. (a) Original rendering of the complete dataset. (b) Clipping planes in the transversal direction. (c) Clipping planes in the sagittal direction. | [
"First, the clinician can use three pairs of clipping planes to interactively cut away uninteresting parts of the volume. Figure 1a shows a rendering of a typical MRA dataset of 100 slices of 256Γ256; the dimensions of the volume are indicated by a bounding box. A major part of the volume is not interesting; we the... | 1 | Clinical evaluation of interactive volume visualization | 1,996 | 10.1109/VISUAL.1996.568134 | http://dx.doi.org/10.1109/VISUAL.1996.568134 | Vis | 100 | null | null | |||
17 | Figure 3: F-actin structures in Dictyostelium amoebae, Resolution 512 Γ484 Γ43 voxels = 10 Mbytes. Upper left a single slice, right surface reconstruction; lower left MIP, right transmission illumination models | [
"Figures 3, 4 and 5 demonstrate the importance of LCM data visualization for detecting unknown structures. In this case, we studied actin filaments in Dictyostelium amoebae. This cell is a model of phagocytosis and chemotaxis, behaving much like the neutrophil leukocytes in our own blood. It also passes through a d... | 3 | Case study: Visualization of laser confocal microscopy datasets | 1,996 | 10.1109/VISUAL.1996.568136 | http://dx.doi.org/10.1109/VISUAL.1996.568136 | Vis | 15 | null | null | |||
18 | Figure 4: F-actin structures. Resolution 512 Γ484 Γ70 voxels = 16.5 Mbytes. Upper left a single slice, right surface reconstruction; lower left MIP, right transmission illumination models | [
"Figures 3, 4 and 5 demonstrate the importance of LCM data visualization for detecting unknown structures. In this case, we studied actin filaments in Dictyostelium amoebae. This cell is a model of phagocytosis and chemotaxis, behaving much like the neutrophil leukocytes in our own blood. It also passes through a d... | 4 | Case study: Visualization of laser confocal microscopy datasets | 1,996 | 10.1109/VISUAL.1996.568136 | http://dx.doi.org/10.1109/VISUAL.1996.568136 | Vis | 100 | null | null | |||
19 | Figure 8: The extra-cellular component of a retina blood vessel of a cat. Resolution of 335 Γ306 Γ67 voxels with a size of 0.16^2 Γ0.2ΞΌm. Upper left a single slice, right surface reconstruction; lower left MIP, right transmission illumination models | [
"Figures 8 and 7 present a microscopic preparation of the tubular structure of a cat retina. The used microscope was able to detect two different fluorophore types separately, thus acquiring two different tissue types simultaneously. The dataset consist of 335 Γ 306 Γ 67 voxels, each with a dimension of 0.16Β² Γ 0.2... | 8 | Case study: Visualization of laser confocal microscopy datasets | 1,996 | 10.1109/VISUAL.1996.568136 | http://dx.doi.org/10.1109/VISUAL.1996.568136 | Vis | 15 | null | null | |||
20 | Figure 4: Wireframe of Chesapeake Bay with cell probes (weighted with respect to time). | [
"Whereas the previous techniques focus on displaying the probes, this technique is better suited to allow comparison between probe data and model output. A cell containing at least one probe datum is rendered as a wireframe, colored with respect to its simulated data value and the colormap. The probe data in a cell... | 4 | Visualization of water quality data for the Chesapeake Bay | 1,996 | 10.1109/VISUAL.1996.568146 | http://dx.doi.org/10.1109/VISUAL.1996.568146 | Vis | 15 | null | null | |||
21 | Figure 6: Cutting Planes and Difference Method (probes weighted with respect to position). | [
"Rather than letting the icon's color represent the function value, we let it represent the (absolute) difference. We compute the difference between the cell-averaged observational data and the model output, divided by the maximum difference. The great advantage of this rendering method is that the user can clearly... | 6 | Visualization of water quality data for the Chesapeake Bay | 1,996 | 10.1109/VISUAL.1996.568146 | http://dx.doi.org/10.1109/VISUAL.1996.568146 | Vis | 15 | null | null | |||
22 | Figure 4: Deformation tensor in a flow past a hemisphere cylinder at incidence | [
"Figure 4 shows hyperstreamlines of the deformation tensor of a flow past a hemisphere-cylinder at a skew angle of incidence. There are two vortex cores in this flow. Two hyperstreamlines along the body are the medium eigenvectors and also define the direction of the vortex core. The upper ring is a minor hyperstre... | 4 | Singularities in nonuniform tensor fields | 1,997 | 10.1109/VISUAL.1997.663857 | http://dx.doi.org/10.1109/VISUAL.1997.663857 | Vis | 15 | null | null | |||
23 | Figure 6: viscous tensor in a flow past a hemisphere cylinder at incidence | [
"Pressure in a stress tensor usually comes from outside forces acting on the medium and it can be disturbingly large compared to the friction related to the viscous flow. This makes it difficult to see the actual effect from the flow itself. The bulk viscosity is the resistance of a compressible flow to the expansi... | 6 | Singularities in nonuniform tensor fields | 1,997 | 10.1109/VISUAL.1997.663857 | http://dx.doi.org/10.1109/VISUAL.1997.663857 | Vis | 15 | null | null | |||
24 | Figure 6: Volume rendering of the Principal Stream Function of Eq.(10) (a) rotate X = 15, Y = 60 (b)rotate 15, Y = -120 (c, d) volume cutting of (a,b) respectively. | [
"This equation is applied to a 10 Γ 10 Γ 10 regular grid. Figure 6 shows the volume-rendered result, using the same rainbow color mapping as in Figure 5. Transparency is set to 0.9 at all points. The value range of the principal stream function is mapped to #5B0 : : : 255#5D for the sake of volume rendering and ima... | 6 | Principal stream surfaces | 1,997 | 10.1109/VISUAL.1997.663859 | http://dx.doi.org/10.1109/VISUAL.1997.663859 | Vis | 100 | null | null | |||
25 | Figure 9: Application of high-pass filter to enhance the principal stream function | [
"This equation is applied to a 10 Γ 10 Γ 10 regular grid. Figure 6 shows the volume-rendered result, using the same rainbow color mapping as in Figure 5. Transparency is set to 0.9 at all points. The value range of the principal stream function is mapped to #5B0 : : : 255#5D for the sake of volume rendering and ima... | 9 | Principal stream surfaces | 1,997 | 10.1109/VISUAL.1997.663859 | http://dx.doi.org/10.1109/VISUAL.1997.663859 | Vis | 15 | null | null | |||
26 | Figure 2: Terrain meshes (Delaunay triangulated) and views rendered at different data resolutions. (a) High resolution: 0.08 triangles/pixel and 1 texels/pixel. (b) Equivalent quality at lower resolution: 0.02 triangles and 0.8 texels/pixels. Note how more DTM points are used in foreground areas or areas of high curvat... | [
"Typical triangulations and rendered images generated by our client system are shown in Fig. 2.",
"A typical DTM is supplied on a regular grid, and this data is usually highly redundant. If the surface is to be approximated by a piecewise-linear 2D function (a collection of planar polygons), a small number of lar... | 2 | Visualization of large terrains in resource-limited computing environments | 1,997 | 10.1109/VISUAL.1997.663863 | http://dx.doi.org/10.1109/VISUAL.1997.663863 | Vis | 100 | null | null | |||
27 | Figure 6: The initial isosurface mesh consisting of 182,376 triangles extracted from 1,646,568 tetrahedra with error=0 | [
"All the images are generated with an isovalue of 55.5. Figs. 6 - 9 correspond to errors: 0.0, 1.5, 2.5 and 4.0 (in voxel size), respectively. Figs. 10 (a) and (b) (see also color pictures) are with errors: 0.0 and 4.0, respectively. Fig. 11 (see also color pictures) is created using different errors: the left part... | 6 | Multiresolution tetrahedral framework for visualizing regular volume data | 1,997 | 10.1109/VISUAL.1997.663869 | http://dx.doi.org/10.1109/VISUAL.1997.663869 | Vis | 15 | null | null | |||
28 | Figure 10: (a) Initial isosurface mesh consisting of 255,256 triangles extracted from 5,390,896 tetrahedra with error=0, (b) Isosurface mesh consisting of 89,300 triangles extracted from 304,468 tetrahedra with error=4.0 (i.e., 94.35% fusion of tetrahedra) | [
"All the images are generated with an isovalue of 55.5. Figs. 6 - 9 correspond to errors: 0.0, 1.5, 2.5 and 4.0 (in voxel size), respectively. Figs. 10 (a) and (b) (see also color pictures) are with errors: 0.0 and 4.0, respectively. Fig. 11 (see also color pictures) is created using different errors: the left part... | 10 | Multiresolution tetrahedral framework for visualizing regular volume data | 1,997 | 10.1109/VISUAL.1997.663869 | http://dx.doi.org/10.1109/VISUAL.1997.663869 | Vis | 100 | null | null | |||
29 | Figure 4. Partial (nearly orthogonal) spherical wavelet reconstruction of a flow field defined over a spherical domain. | [
"More interestingly, we have applied our new wavelets to data consisting of a vector field defined over the sphere. In the example of Figure 4, we took a known vector field function and evaluated it to obtain the data. We applied the wavelet of equation (8). In the left column we show three views of the reconstruct... | 4 | Haar wavelets over triangular domains with applications to multiresolution models for flow over a sphere | 1,997 | 10.1109/VISUAL.1997.663871 | http://dx.doi.org/10.1109/VISUAL.1997.663871 | Vis | 100 | null | null | |||
30 | Figure 5. Spherical wavelet reconstructed of wind data from global weather model | [
"More interestingly, we have applied our new wavelets to data consisting of a vector field defined over the sphere. In the example of Figure 4, we took a known vector field function and evaluated it to obtain the data. We applied the wavelet of equation (8). In the left column we show three views of the reconstruct... | 5 | Haar wavelets over triangular domains with applications to multiresolution models for flow over a sphere | 1,997 | 10.1109/VISUAL.1997.663871 | http://dx.doi.org/10.1109/VISUAL.1997.663871 | Vis | 100 | null | null | |||
31 | Figure 2. Subdivision of a slice of CFS data. The significant regions-of-interest are marked with arrows. (a) Mask driven quadtree decomposition of original slice. (b) Mask. The mask has the highest value in the marked regions, which are regions of significant structures. | [
"Structure Detection - We use a wavelet based analysis algorithm that locates significant structures which persist across the scales of the wavelet decomposition. As a result of the induced subdivision of the frequency spectrum of the image, the image is now decomposed into a pyramid consisting of bandpass images. ... | 2 | Wavelet-based multiresolutional representation of computational field simulation datasets | 1,997 | 10.1109/VISUAL.1997.663872 | http://dx.doi.org/10.1109/VISUAL.1997.663872 | Vis | 100 | null | null | |||
32 | Figure 3: A scaling field that could be used, in regions of value greater than unity, to magnify the screen distance traversed by a unit mouse motion; similarly, in regions of value less than unity, this field would slow the mouse response to provide fine-grained control in those limited areas where it is required. | [
"Sensitivity Fields. A number of applications have identifiable areas where one wants to have very fine control, and others where one wants coarse control for quickly traversing large, uninteresting areas. We note two examples that fit cleanly into our framework: (1) Velocity-based displacement. Several common mous... | 3 | Constrained 3D navigation with 2D controllers | 1,997 | 10.1109/VISUAL.1997.663876 | http://dx.doi.org/10.1109/VISUAL.1997.663876 | Vis | 15 | null | null | |||
33 | Figure 6: Camera path constrained to plane with camera orientation modulated by terrain gradient. (a) View of path and camera model control points on constraint surface. (b) View using camera model field at selected point. | [
"Wandering Camera Path with Wandering View. In a traditional computer animation, the camera itself may follow many different constraints such as looking at a single point on the ground throughout the motion, tracking a moving object in the scene, or staring in a fixed direction. Figure 5(a,b) shows a generalization... | 6 | Constrained 3D navigation with 2D controllers | 1,997 | 10.1109/VISUAL.1997.663876 | http://dx.doi.org/10.1109/VISUAL.1997.663876 | Vis | 15 | null | null | |||
34 | Figure 7: Camera path constrained to complex surface with camera orientation keyed to constraint surface normal and modulated by terrain gradient. (a) View of path and camera model control points on constraint surface. (b) View using camera model field at selected point. | [
"Modulation by Data. We can immediately go beyond the already useful idea of having predetermined camera parameters at each point of the navigable space by defining modifiers of the default parameters. In Figure 7, we show the result of using the gradient βΟ of the terrain elevation model as a cue: starting with an... | 7 | Constrained 3D navigation with 2D controllers | 1,997 | 10.1109/VISUAL.1997.663876 | http://dx.doi.org/10.1109/VISUAL.1997.663876 | Vis | 15 | null | null | |||
35 | Figure 8: Spotlight focused on an area of interest that is slightly displaced from camera gaze and motion directions. This allows greater flexibility in keeping the context while redirecting attention. | [
"Fog, Spotlights, etc. The actual scene appearance can equally well be modulated to suit the designer's needs. We suggest the following methods: (1) Fog. As one passes through a scene, one can limit the visibility to a handful of key regions by obscuring the most distant objects. Other application-dependent depth c... | 8 | Constrained 3D navigation with 2D controllers | 1,997 | 10.1109/VISUAL.1997.663876 | http://dx.doi.org/10.1109/VISUAL.1997.663876 | Vis | 15 | null | null | |||
36 | Figure 11: (a) Example of a multiple-valued constraint configuration. (b) View from marked point first time around the path. (c) View from marked point second time around the path, showing a different detail to the viewer. | [
"Multiple Coverings. Another fundamental technique is the 'multiple covering' navigation surface. (Readers with mathematical backgrounds will recognize this as a relative of Riemann surfaces in complex variable theory.) Here, one creates a surface that may come back to the same point by many different routes; a sim... | 11 | Constrained 3D navigation with 2D controllers | 1,997 | 10.1109/VISUAL.1997.663876 | http://dx.doi.org/10.1109/VISUAL.1997.663876 | Vis | 100 | null | null | |||
37 | Figure 8 - Turbine shell shown at various reduction levels. Shell has many features requiring vertex splitting. | [
"Color Plates 3(a)-(f) show results for the heat exchanger. Figures 8(a)-(f) show results for the turbine shell. Figures 9(a)-(f) show the results for the turbine blade. Note that in each case the onset of vertex splitting is shown. In some of these color plates the red edges are used to indicate mesh boundaries, w... | 8 | A topology modifying progressive decimation algorithm | 1,997 | 10.1109/VISUAL.1997.663883 | http://dx.doi.org/10.1109/VISUAL.1997.663883 | Vis | 15 | null | null | |||
38 | Figure 1: This sample screen shows a number of different visualizations approximately when the tornado touched down. Shown are an isosurface over rain-water density, and particle flow from three emitters, at the ground plane, at the base of the funnel, and vertically along the column of the funnel. | [
"CAVEvis was designed originally to explore data sets generated from the simulation of severe thunderstorms and tornados [9]. The tornado simulations generated 40 gigabytes of data consisting of wind velocity vector fields and scalar fields for temperature, pressure, water density, and vorticity, over hundreds of t... | 1 | CAVEvis: distributed real-time visualization of time-varying scalar and vector fields using the CAVE virtual reality theater | 1,997 | 10.1109/VISUAL.1997.663896 | http://dx.doi.org/10.1109/VISUAL.1997.663896 | Vis | 15 | null | null | |||
39 | Figure 3: LIC image (a), OLIC images for two flows with opposite orientation (b), (c) | [
"In figure 3 the difference between LIC and OLIC is clearly visible. Figure 3(a) shows the LIC image of a circular flow but it is not recognizable if the flow is in clockwise or counterclockwise orientation. Figure 3(b) shows the OLIC image of a circular clockwise flow and figure 3(c) shows the OLIC image of a circ... | 3 | Fast oriented line integral convolution for vector field visualization via the Internet | 1,997 | 10.1109/VISUAL.1997.663897 | http://dx.doi.org/10.1109/VISUAL.1997.663897 | Vis | 15 | null | null | |||
40 | Figure 6: A ray-cast image of the head data set, using the X-ray-like opacity model. | [
"Tocomparethe rendered images from voxel-by-voxel ray casting and from the compression domain rendering algorithm, we render the head data set using the two algorithms under different opacity transfer functions. Figure 6 and 9 show the rendered images using ray casting and compression domain rendering, assuming tha... | 6 | Integrated volume compression and visualization | 1,997 | 10.1109/VISUAL.1997.663900 | http://dx.doi.org/10.1109/VISUAL.1997.663900 | Vis | 15 | null | null | |||
41 | Figure 6: (a) Range data of a head, (b) data and initialized model, (c) intermediate stage of evolution and (d) the fitted model polygon. | [
"In Fig.6, we present shape recovery from range data using our model. The model, with 96 faces and 98 vertices, 8 of them being extraordinary vertices of valence 3, was initialized inside a set of 1,779 range data points. The final fitted model has a control polygon of 384 faces with 386 vertices and thus the shape... | 6 | Dynamic smooth subdivision surfaces for data visualization | 1,997 | 10.1109/VISUAL.1997.663905 | http://dx.doi.org/10.1109/VISUAL.1997.663905 | Vis | 15 | null | null | |||
42 | Figure 7: Image plate showing the results of iso-surface extraction and volume ray-casting | [
"As we can see in Table 2 the time needed to generate the triangles is decreasing significantly compared to the standard marching cubes algorithm. Looking especially at Head 1 we see, that the number of triangles is about one half and the processing time is about 6 times faster. In the case of the Abdomen 3 (Fig. 7... | 7 | The multilevel finite element method for adaptive mesh optimization and visualization of volume data | 1,997 | 10.1109/VISUAL.1997.663907 | http://dx.doi.org/10.1109/VISUAL.1997.663907 | Vis | 15 | null | null | |||
43 | Figure 2: An oblique view of the flow, rendered with (right) and without (left) visibility-impeding 3D halos. | [
"Because we are using volume rendering for this application, and the presence of an ' occluding element' will be indicated by a continuous rather than a binary function, it is most appropriate to use a similarly continuous metric to define the locations and strengths of the gaps that will mark the depth discontinui... | 2 | Strategies for effectively visualizing 3D flow with volume LIC | 1,997 | 10.1109/VISUAL.1997.663912 | http://dx.doi.org/10.1109/VISUAL.1997.663912 | Vis | 100 | null | null | |||
44 | Color Plate 1: A volume rendered image of a 3D flow dataset generated using volume LIC. Color variations are used to differentiate the individual streamlines and to facilitate the conscious direction of focused attention to subsets of these lines. Depth discontinuities are explicitly indicated by means of a locally-def... | [
"By assigning each of the different input texture elements one of a small number of harmonious yet readily distinguishable different hues, as demonstrated in color plate 1, it may also be possible to facilitate the focusing of selective attention on different subsets of the LIC-generated lines in turn. Colors can b... | null | Strategies for effectively visualizing 3D flow with volume LIC | 1,997 | 10.1109/VISUAL.1997.663912 | http://dx.doi.org/10.1109/VISUAL.1997.663912 | Vis | 100 | null | null | |||
45 | Color Plate 2: Local depth order relationships are difficult to appreciate when continuity information is lacking, even if halos and color variations are used. | [
"Recent research [ 13, 231 indicates that our understanding of surface shape may be derived not from an aggregation of individual estimates of the surface normal directions at distributed points, but from the organization of space defined by local depth order relationships between adjoining regions. It follows that... | null | Strategies for effectively visualizing 3D flow with volume LIC | 1,997 | 10.1109/VISUAL.1997.663912 | http://dx.doi.org/10.1109/VISUAL.1997.663912 | Vis | 100 | null | null | |||
46 | Color Plate 3: When streamlines are represented as a volume texture, color and opacity information can be easily redefined based on any function of the flow values across the volume. In this image, vorticity magnitude is locally mapped to streamline color using a heated-object colorscale. The striations along the axial... | [
"When streamlines are made available as a scan-converted volume texture, one has the option of easily employing standard volume rendering procedures to isolate specific regions of interest or to convey information about related scalar quantities over the flow, as shown in color plate 3, by modifying the local color... | null | Strategies for effectively visualizing 3D flow with volume LIC | 1,997 | 10.1109/VISUAL.1997.663912 | http://dx.doi.org/10.1109/VISUAL.1997.663912 | Vis | 100 | null | null | |||
47 | Fig 1: A 12-block configuration used for the simulation of an Electrostatic Precipitator. (Data courtesy of W. Egli, ABB Corp. Research Ltd. Switzerland) | [
"As hardware vendors provide scientists with bigger machines, very large datasets, single-block or multi-block (Fig. 1) are becoming ubiquitous. They present many challenges to scientific visualization softwares, but many opportunities exist to foster the efficient management of visualization resources."
] | 1 | Towards efficient visualization support for single-block and multi-block datasets | 1,997 | 10.1109/VISUAL.1997.663913 | http://dx.doi.org/10.1109/VISUAL.1997.663913 | Vis | 15 | null | null | |||
48 | Figure 3: Qualitative brushing with multiple resolutions. See also Color Plate 1. | [
"A qualitative brushing process highlights interesting areas according to the values of the data. For example, if the most important part of the figure is the skull, only the data values corresponding to bones and joints are used to generate the isosurfaces. The rest of the data, mainly flesh and skin, can be obtai... | 3 | Brushing techniques for exploring volume datasets | 1,997 | 10.1109/VISUAL.1997.663914 | http://dx.doi.org/10.1109/VISUAL.1997.663914 | Vis | 15 | null | null | |||
49 | Figure 5: Volume brushing of the jaw bone area. See also Color Plate 3. | [
"A volume brush is a high resolution sub-volume within a coarse volume dataset. Its goal is to brush smaller regions of data values which cannot be achieved solely by a qualitative or a planar brush. For example, the jaw bone area of the skull, which is totally blocked by the forehead in Figure 3, can easily be see... | 5 | Brushing techniques for exploring volume datasets | 1,997 | 10.1109/VISUAL.1997.663914 | http://dx.doi.org/10.1109/VISUAL.1997.663914 | Vis | 15 | null | null | |||
50 | Figure 1: Three LODs for a CAD model including (a) an 18-sided convex hull, (b) a decimated representation, and (c) the full resolution model. | [
"Figure 1 shows three levels of resolution for a geometric model produced by a CAD system. The convex hull representation has 18 polygons, the decimated version generated with a desired reduction rate of 95% has 15,718 triangles, and the full resolution model has 314,393 triangles.",
"The full resolution CAD mode... | 1 | Interactive visualization of aircraft and power generation engines | 1,997 | 10.1109/VISUAL.1997.663927 | http://dx.doi.org/10.1109/VISUAL.1997.663927 | Vis | 100 | null | null | |||
51 | Figure 4: A complex CAD environment shown at (a) full resolution and viewed on (b) an InfiniteReality, (c) an Octane, and (d) an O^2 using the same desired frame rate. | [
"Using Galileo on a wide range of workstations we found that the system generally ' feels' the same but ' looks' different across the different platforms. The frame rate, which is controlled by the user, is nearly independent of the underlying graphics hardware. The LODs that are selected to be rendered at the give... | 4 | Interactive visualization of aircraft and power generation engines | 1,997 | 10.1109/VISUAL.1997.663927 | http://dx.doi.org/10.1109/VISUAL.1997.663927 | Vis | 100 | null | null | |||
52 | Figure 2: The pieces of this CAD model are independently colored to show the organization of the polygons. | [
"The full resolution CAD model shown in Figure 1 is composed of 88 pieces with an average of 3573 triangles in each. These pieces are independently colored in Figure 2 in order show the organization of the polygons. The division of parts into pieces is helpful not only in rendering methods, but also in computationa... | 2 | Interactive visualization of aircraft and power generation engines | 1,997 | 10.1109/VISUAL.1997.663927 | http://dx.doi.org/10.1109/VISUAL.1997.663927 | Vis | 100 | null | null | |||
53 | Figure 1: Mapping of strain values using plain color mapping (up-per picture) and using iso-contours (lower picture). | [
"Another approach for the use of one-dimensional textures is the creation of iso-contours. Using a texture with less colors and sharp borders between the colors automatically creates iso-contours on the textured model without any calculation of intersection points between the contour edges and the polygons of the m... | 1 | Case study: efficient visualization of physical and structural properties in crash-worthiness simulations | 1,997 | 10.1109/VISUAL.1997.663928 | http://dx.doi.org/10.1109/VISUAL.1997.663928 | Vis | 15 | null | null | |||
54 | Figure 8: Intrusion of the car body during a side impact collision | [
"If the acceptable limit is changed or the intrusion has to be investigated using a broader or tighter range of interest, no additional computational effort is necessary, only the texture definition has to be adapted. The use of iso-contouring showing the intrusion of the car body during a side impact collision is ... | 8 | Case study: efficient visualization of physical and structural properties in crash-worthiness simulations | 1,997 | 10.1109/VISUAL.1997.663928 | http://dx.doi.org/10.1109/VISUAL.1997.663928 | Vis | 100 | null | null | |||
55 | Figure 12: Force tube depicting the force flux through a longitudinal structure (20 milliseconds after the crash) | [
"The centers of gravity of the force tube are defined by a line parallel to the trace line of the structure. The distance of the tube to the trace-line can be chosen by the user. Thus, the tube can be positioned next to the longitudinal structure with a reasonable spacing. Several cylinders are positioned around th... | 12 | Case study: efficient visualization of physical and structural properties in crash-worthiness simulations | 1,997 | 10.1109/VISUAL.1997.663928 | http://dx.doi.org/10.1109/VISUAL.1997.663928 | Vis | 100 | null | null | |||
56 | Figure 9: A surface with three branches. (a) interpolated contours and slope line(b) perspective view displayed by polygonal patches. | [
"To confirm the capabilities of the proposed method, we display two kinds of complicated models. Fig. 8 illustrates a surface with two branches, in which the default gradient conditions are set on the boundary contours (thick lines). The interpolated sub-contours and the slope lines are given in Fig. 8(a) and the r... | 9 | Contour interpolation and surface reconstruction of smooth terrain models | 1,998 | 10.1109/VISUAL.1998.745281 | http://dx.doi.org/10.1109/VISUAL.1998.745281 | Vis | 15 | null | null | |||
57 | Figure 11: A surface reconstructed from a part of a real map. (a) the surface reconstructed without consideration of smoothness across the contours. (b) the surface reconstructed by the proposed method. (c) the interpolated contours and slope lines of the surface in (b). | [
"To confirm the capabilities of the proposed method, we display two kinds of complicated models. Fig. 8 illustrates a surface with two branches, in which the default gradient conditions are set on the boundary contours (thick lines). The interpolated sub-contours and the slope lines are given in Fig. 8(a) and the r... | 11 | Contour interpolation and surface reconstruction of smooth terrain models | 1,998 | 10.1109/VISUAL.1998.745281 | http://dx.doi.org/10.1109/VISUAL.1998.745281 | Vis | 15 | null | null | |||
58 | Figure 10: A surface reconstructed from two contours with very different shapes from each other. (a) interpolated contours and slope lines. (b) perspective view displayed by polygonal patches. | [
"To confirm the capabilities of the proposed method, we display two kinds of complicated models. Fig. 8 illustrates a surface with two branches, in which the default gradient conditions are set on the boundary contours (thick lines). The interpolated sub-contours and the slope lines are given in Fig. 8(a) and the r... | 10 | Contour interpolation and surface reconstruction of smooth terrain models | 1,998 | 10.1109/VISUAL.1998.745281 | http://dx.doi.org/10.1109/VISUAL.1998.745281 | Vis | 15 | null | null | |||
59 | Figure 1: Scalar topology examples (a) Visualization of topology and density in a pion collision simulation (b) same image with isocontours (c-d) Visualization of a scalar-valued mathematical function (e) Scalar topology diagram of the 3D wave f computed for a high potential iron protein. (f) Close-up of volume rendere... | [
"Isocontours, or constant valued curves and surfaces from continuous 2D and 3D scalar fields, are a common visualization technique for displaying scalar field structure[21]. By their definition, isocontoursrepresent the data only at discrete computing integral curves in the gradient field from saddle points to an a... | 1 | Visualization of scalar topology for structural enhancement | 1,998 | 10.1109/VISUAL.1998.745284 | http://dx.doi.org/10.1109/VISUAL.1998.745284 | Vis | 100 | null | null | |||
60 | Figure 6: Enhanced visualization of MM5 simulation with feature tracking information. The top images are four steps from the simulation. The middle image is one set of features extracted. The bottom graphs contains quantifications of the large object. | [
"In Figure 6, the feature tracking algorithm was applied to the cloud water simulation (regular datasets). This is a simulation using the MM5 modeling system of cloud formation over the eastern United States. The simulation consists of 25 datasets at a resolution of 35x41x23. Features are first extracted at the thr... | 6 | Tracking scalar features in unstructured datasets | 1,998 | 10.1109/VISUAL.1998.745288 | http://dx.doi.org/10.1109/VISUAL.1998.745288 | Vis | 15 | null | null | |||
61 | Figure 1: Airfoil dataset in physical space (left) and velocity component space (right). The edges of the physical space image are axis-aligned, with x increasing to the right, and y increasing to the top. The vertices of the computational mesh in velocity component space are located by the x and y components of their ... | [
"Figure 1 shows a CFD dataset plotted in both physical space and in velocity component space . The original dataset features a curvilinear mesh fitted to an airfoil, and the mesh exhibits increased sampling density near the surface, and above, below, and behind the trailing edge of the wing. The mesh actually exten... | 1 | Feature detection in linked derived spaces | 1,998 | 10.1109/VISUAL.1998.745289 | http://dx.doi.org/10.1109/VISUAL.1998.745289 | Vis | 100 | null | null | |||
62 | Figure 2: Selections in velocity component space (left) with corresponding preimages in physical space (right). Unselected regions are shown in gray; selected regions and preimage in black. | [
"Figure 2 shows the same pair of images as Figure 1, but in reversed order. In the left image, we have selected those points whose velocity magnitude exceeds that of the freestream by some amount β in velocity component space this just amounts to selecting points that are farther from the origin than the image of t... | 2 | Feature detection in linked derived spaces | 1,998 | 10.1109/VISUAL.1998.745289 | http://dx.doi.org/10.1109/VISUAL.1998.745289 | Vis | 15 | null | null | |||
63 | Figure 3: Evolution of recirculation bubble in velocity component space. See accompanying video for live version. | [
"Viewing these dynamic portraits in certain spaces can lead to an understanding of the field behavior which may be difficult or impossible to discern in the base space. For example, Figure 3 shows several images from an oscillating airfoil at various angles of attack. As the angle of attack increases, the airfoil a... | 3 | Feature detection in linked derived spaces | 1,998 | 10.1109/VISUAL.1998.745289 | http://dx.doi.org/10.1109/VISUAL.1998.745289 | Vis | 15 | null | null | |||
64 | Figure 1: Groups of paper strips are used to form a pexel that supports variation of the three perceptual texture dimensions (height, density and randomness): (a) each pexel has one of its dimensions varied across three discrete values; (b) a map of North America, pexels represent areas of high cultivation, height mapp... | [
"In order to support variation of height, density, and regularity, we built pexels that look like a collection of paper strips. The user maps attributes in the dataset to the density (which controls the number of strips in a pexel), height, and regularity of each pexel. Unlike Gabor filters or Wold features, which ... | 1 | Building perceptual textures to visualize multidimensional datasets | 1,998 | 10.1109/VISUAL.1998.745292 | http://dx.doi.org/10.1109/VISUAL.1998.745292 | Vis | 15 | null | null | |||
65 | Figure 2: Three displays of pexels with different regularity and a 5 Γ3 patch from the corresponding autocorrelation graphs: (a) a completely regular display, resulting in sharp peaks of height 1.0 at regular intervals in the autocorrelation graph; (b) a display with irregularly-spaced pexels, peaks in the graph are re... | [
"As a practical example, consider Figure 2a (pexels on a regular underlying grid), Figure 2b (pexels on an irregular grid), and Figure 2c (pexels on a random grid). Autocorrelation was computed on the orthogonal projection of each image. A 5 Γ 3 patch from the center of the corresponding autocorrelation graph is sh... | 2 | Building perceptual textures to visualize multidimensional datasets | 1,998 | 10.1109/VISUAL.1998.745292 | http://dx.doi.org/10.1109/VISUAL.1998.745292 | Vis | 15 | null | null | |||
66 | Figure 5: Two displays with a regular target, both displays should be compared with the target shown in Figure 3b: (a) larger target, an 8 Γ8 target in a sea of sparse pexels; (b) denser background, a 2 Γ2 target in a sea of dense pexels (target group located below and left of center) | [
"One way to make regularity targets easier to identify is by increasing the size of the target patch. Figure 5a shows an 8 Γ 8 regular target in a sea of random pexels. This target is much easier to find, compared to the 2 Γ 2 patch shown in Figure 3b. Unfortunately, we cannot guarantee that the values in a dataset... | 5 | Building perceptual textures to visualize multidimensional datasets | 1,998 | 10.1109/VISUAL.1998.745292 | http://dx.doi.org/10.1109/VISUAL.1998.745292 | Vis | 100 | null | null | |||
67 | Figure 1: Left: The user view, Right: The same isosurface from a 90 degree angle to the user view, illustrating the incomplete reconstruction. | [
"In this paper we present a novel view dependent isosurface extraction approach, as illustrated in Figure 1, which further reduces the search, construction and display by only visiting the cells that contain the visible portion of the isosurface from a given view point. Our approach is based on a hierarchical front... | 1 | View dependent isosurface extraction | 1,998 | 10.1109/VISUAL.1998.745300 | http://dx.doi.org/10.1109/VISUAL.1998.745300 | Vis | 15 | null | null | |||
68 | Figure 2: Extracted isosurface. A cut plane through the full and view-dependent isosurfaces extracted from the same view point as in Figure 1. Note the large internal structures that are part of the full isosurface but are not part of the view-dependent isosurface | [
"the dataset with dynamic pruning of sections that are hidden from the view point by previously extracted sections of the isosurface. Figure 2 shows the potential saving of such an approach. Note the large section of the isosurface, which represents the internal organs in the head, yet is not part of the view-depen... | 2 | View dependent isosurface extraction | 1,998 | 10.1109/VISUAL.1998.745300 | http://dx.doi.org/10.1109/VISUAL.1998.745300 | Vis | 15 | null | null | |||
69 | Figure 13: Visualizations of Greyscale World Map Data Using Two Different Resolutions | [
"Our first comparison uses a greyscale world data map with simulated points distributed over the surface of dry land. Figure 13 shows the result of visualizing the data using our nearest-neighbor, curve-based, and Gridfit algorithms on two different resolutions. The visualizations clearly shows their advantages and... | 13 | The Gridfit algorithm: an efficient and effective approach to visualizing large amounts of spatial data | 1,998 | 10.1109/VISUAL.1998.745301 | http://dx.doi.org/10.1109/VISUAL.1998.745301 | Vis | 15 | null | null | |||
70 | Figure 5: βWaveβ for Buddhist article using MRLs 6 β 8. See also CP2. | [
"In Figure 3, each document subchunk is depicted as a geometric object whose centroid coordinates are provided by the MDS calculation and whose base radius is determined by a measure of variability across the cylinders. One variability candidate could be the length of each document subchunk.",
"The visualization ... | 5 | TOPIC ISLANDS TM - a wavelet-based text visualization system | 1,998 | 10.1109/VISUAL.1998.745302 | http://dx.doi.org/10.1109/VISUAL.1998.745302 | Vis | 100 | null | null | |||
71 | Figure 2: Result of simplifying a bunny model. Only 1.4% of the original faces remain. Centered around each vertex is an isosurface of the corresponding quadric. | [
"We have also suggested that quadrics characterize the local shape of the surface. This is apparent in Figure 2, which illustrates the quadric isosurfaces produced by the simplification of a bunny model. For vertices on creases, such as on the neck and ears, the ellipsoids are cigar shaped. They are elongated in th... | 2 | Simplifying surfaces with color and texture using quadric error metrics | 1,998 | 10.1109/VISUAL.1998.745312 | http://dx.doi.org/10.1109/VISUAL.1998.745312 | Vis | 15 | null | null | |||
72 | Figure 7: At left: a curved surface (18,050 faces) with colors at each vertex. At right: 1,000 face approximation. Notice that mesh edges follow the color contours. | [
"Boundaries may also occur in discrete surface attributes. Consider a surface similar to the one pictured in Figure 7 where each face would be assigned a color from a small discrete palette. Or perhaps we have a map 4-colored by country. Each edge dividing two faces of different colors can be marked as a boundary. ... | 7 | Simplifying surfaces with color and texture using quadric error metrics | 1,998 | 10.1109/VISUAL.1998.745312 | http://dx.doi.org/10.1109/VISUAL.1998.745312 | Vis | 100 | null | null | |||
73 | Figure 8: Simplifying geometry only: A very complex model of 1,085,634 faces (a) is simplified to 20,000 faces (bβc) and 1,000 faces (dβe). | [
"Figures 8a-e illustrate the performance of the fundamental algorithm on a very complex surface which is purely geometric. The original model contains 1,085,634 faces; the approximations shown contain 20,000 and 1,000 faces. Producing these approximations might require hours with some algorithms. However, on a Pent... | 8 | Simplifying surfaces with color and texture using quadric error metrics | 1,998 | 10.1109/VISUAL.1998.745312 | http://dx.doi.org/10.1109/VISUAL.1998.745312 | Vis | 15 | null | null | |||
74 | Figure 7: A piecewise-linear example: The values of the scalar field are only rendered on the boundary of the grid. | [
"Results of our work are shown in Figures 7-15. We utilize voxelized data sets, where each voxel of the original data set is initially split into six tetrahedra (see [14]). We utilized the 'greedy algorithm of Section 4.3 in each case, and specified a maximum error for each approximating mesh.",
"Our first exampl... | 7 | Simplification of tetrahedral meshes | 1,998 | 10.1109/VISUAL.1998.745315 | http://dx.doi.org/10.1109/VISUAL.1998.745315 | Vis | 100 | null | null | |||
75 | Figure 10: An isosurface of a skull data set containing 2.7 million tetrahedra in the original mesh. | [
"Results of our work are shown in Figures 7-15. We utilize voxelized data sets, where each voxel of the original data set is initially split into six tetrahedra (see [14]). We utilized the βgreedyβ algorithm of Section 4.3 in each case, and specified a maximum error for each approximating mesh.",
"Our second exam... | 10 | Simplification of tetrahedral meshes | 1,998 | 10.1109/VISUAL.1998.745315 | http://dx.doi.org/10.1109/VISUAL.1998.745315 | Vis | 15 | null | null | |||
76 | Figure 11: Simplified skull data set: An error bound was chosen that reduces the data set to approximately 910,000 tetrahedra. The same isosurface value was used as in Figure 10. | [
"Results of our work are shown in Figures 7-15. We utilize voxelized data sets, where each voxel of the original data set is initially split into six tetrahedra (see [14]). We utilized the 'greedy algorithm of Section 4.3 in each case, and specified a maximum error for each approximating mesh.",
"Our second examp... | 11 | Simplification of tetrahedral meshes | 1,998 | 10.1109/VISUAL.1998.745315 | http://dx.doi.org/10.1109/VISUAL.1998.745315 | Vis | 15 | null | null | |||
77 | Figure 8: The underlying tetrahedral mesh for the function shown in Figure 7. The field contains 41,154 tetrahedra. The edges are colored according to the scalar function. | [
"Results of our work are shown in Figures 7-15. We utilize voxelized data sets, where each voxel of the original data set is initially split into six tetrahedra (see [14]). We utilized the 'greedy algorithm of Section 4.3 in each case, and specified a maximum error for each approximating mesh.",
"Our first exampl... | 8 | Simplification of tetrahedral meshes | 1,998 | 10.1109/VISUAL.1998.745315 | http://dx.doi.org/10.1109/VISUAL.1998.745315 | Vis | 15 | null | null | |||
78 | Figure 12: Skull data simplification containing approximately 209,000 tetrahedra: The isosurface is shown using the same value as in Figure 10. | [
"Results of our work are shown in Figures 7-15. We utilize voxelized data sets, where each voxel of the original data set is initially split into six tetrahedra (see [14]). We utilized the βgreedy algorithm of Section 4.3 in each case, and specified a maximum error for each approximating mesh.",
"Our second examp... | 12 | Simplification of tetrahedral meshes | 1,998 | 10.1109/VISUAL.1998.745315 | http://dx.doi.org/10.1109/VISUAL.1998.745315 | Vis | 15 | null | null | |||
79 | Figure 13: A ray-traced image of a section of the brain of a Macaque monkey. The original data set contains approximately 1.3 million tetrahedra. | [
"Results of our work are shown in Figures 7-15. We utilize voxelized data sets, where each voxel of the original data set is initially split into six tetrahedra (see [14]). We utilized the 'greedy algorithm of Section 4.3 in each case, and specified a maximum error for each approximating mesh.",
"The third exampl... | 13 | Simplification of tetrahedral meshes | 1,998 | 10.1109/VISUAL.1998.745315 | http://dx.doi.org/10.1109/VISUAL.1998.745315 | Vis | 15 | null | null | |||
80 | Figure 15: A ray-traced image of a simplified brain data set. An error bound was chosen that reduced the set to approximately 158,000 tetrahedra. | [
"Results of our work are shown in Figures 7-15. We utilize voxelized data sets, where each voxel of the original data set is initially split into six tetrahedra (see [14]). We utilized the βgreedyβ algorithm of Section 4.3 in each case, and specified a maximum error for each approximating mesh.",
"The third examp... | 15 | Simplification of tetrahedral meshes | 1,998 | 10.1109/VISUAL.1998.745315 | http://dx.doi.org/10.1109/VISUAL.1998.745315 | Vis | 15 | null | null | |||
81 | Figure 14: A ray-traced image of the simplified brain data set. An error bound was chosen that reduces the data set to approximately 700,000 tetrahedra. | [
"Results of our work are shown in Figures 7-15. We utilize voxelized data sets, where each voxel of the original data set is initially split into six tetrahedra (see [14]). We utilized the 'greedy algorithm of Section 4.3 in each case, and specified a maximum error for each approximating mesh.",
"The third exampl... | 14 | Simplification of tetrahedral meshes | 1,998 | 10.1109/VISUAL.1998.745315 | http://dx.doi.org/10.1109/VISUAL.1998.745315 | Vis | 15 | null | null | |||
82 | Figure 2: Single point load Boussinesq problem - Deviator/Isotropic comparison on x-axis slice | [
"Figures 1 to 3 and Color Plates 1 through 5 show visualizations produced with our techniques on a well understood analytical data set and more complicated 'real world' data sets generated from CFD simulations.",
"Figures 1 to 3 and Color Plate 1 show verification tests performed using a single point load Boussin... | 2 | Interactive deformations from tensor fields | 1,998 | 10.1109/VISUAL.1998.745316 | http://dx.doi.org/10.1109/VISUAL.1998.745316 | Vis | 15 | null | null | |||
83 | Figure 4: Possible filter shapes for continuous directional convolution, one-dimensional filter, triangle swept along a streamline, swept rectangle, βspotβ, and low-pass filter. | [
"The shape of the filter depends on the desired effect. Possibilities are a line with or without a certain width or a more complicated two-dimensional shape (see Figure 4). In terms of frequencies, the purpose of the filter is to achieve an anisotropic filtering of the input where the maximum frequency in the field... | 4 | Comparing LIC and spot noise | 1,998 | 10.1109/VISUAL.1998.745324 | http://dx.doi.org/10.1109/VISUAL.1998.745324 | Vis | 100 | null | null | |||
84 | Figure 8: LIC (left) and spot noise (right) images of turbulent flow around a block. | [
"Among the techniques used for the visualization of vector fields, texture based methods are a recent development. By using texture, a continuous visualization of a two-dimensional vector field can be presented. Figure 8 shows a visualization of a slice from a direct numerical simulation using texture. The images s... | 8 | Comparing LIC and spot noise | 1,998 | 10.1109/VISUAL.1998.745324 | http://dx.doi.org/10.1109/VISUAL.1998.745324 | Vis | 100 | null | null | |||
85 | Figure 9: LIC (left) and spot noise (right) images with equal pixel coverage using a filter length of 20 and a spot radius of 0.005 (top) and a filter length of 40 and a spot radius of 0.007 (bottom). | [
"Because the large majority of grid cells in the data is smaller than a texel the amount of detail which could be visible is limited by the texture resolution. For further investigation we used a detail behind the block. This is shown in In Figure 9. In the lower part of the images the block is visible. The data re... | 9 | Comparing LIC and spot noise | 1,998 | 10.1109/VISUAL.1998.745324 | http://dx.doi.org/10.1109/VISUAL.1998.745324 | Vis | 100 | null | null | |||
86 | Figure 1: The (3,5) torus knot, a complex periodic 3D curve. (a) The line drawing is nearly useless as a 3D representation. (b) A tubing based on parallel transporting an initial reference frame produces an informative visualization, but is not periodic. (c) The arrow in this closeup exposes the subtle but crucial non-... | [
"Figure 1 summarizes the basic class of problems involving curves that will concern us here. The line drawing (a) of a (3,5) torus knot provides no useful information about the 3D structure. Improving the visualization by creating a tubing involves a subtle dilemma that we attempt to expose in the rest of the figur... | 1 | Constrained optimal framings of curves and surfaces using quaternion Gauss maps | 1,998 | 10.1109/VISUAL.1998.745326 | http://dx.doi.org/10.1109/VISUAL.1998.745326 | Vis | 15 | null | null | |||
87 | Figure 2: (a) A smooth 3D surface patch having a non-orthogonal parameterization, along with its geometrically-fixed normals at the four corners. No unique orthonormal frame is derivable from the parameterization. If we imitate parallel transport for curves to evolve the initial frame at the top corner to choose the fr... | [
"Figure 2 illustrates a corresponding problem for surface patches. While the normals to the four corners of the patch are always welldefined (a), one finds two different frames for the bottom corner depending upon whether one parallel transports the initial frame around the left-hand path (b) or the right-hand path... | 2 | Constrained optimal framings of curves and surfaces using quaternion Gauss maps | 1,998 | 10.1109/VISUAL.1998.745326 | http://dx.doi.org/10.1109/VISUAL.1998.745326 | Vis | 100 | null | null | |||
88 | Figure 6: (a) A trefoil torus knot. (b) Its quaternion Frenet frame projected to 3D. For this trefoil knot, the frame does not close on itself in quaternion space unless the curve is traversed twice, corresponding to the double-valued βmirrorβ image of the rotation space that can occur in the quaternion representation.... | [
"Closed Curve Example. In Figure 6, we show a simple closed curve, the trefoil knot, the quaternion plot of its periodic Frenet frame, and, just to show we can do it, the entire constraint surface in which the Frenet frame and all other possible quaternion framings of the trefoil must lie. In the next section, we s... | 6 | Constrained optimal framings of curves and surfaces using quaternion Gauss maps | 1,998 | 10.1109/VISUAL.1998.745326 | http://dx.doi.org/10.1109/VISUAL.1998.745326 | Vis | 100 | null | null | |||
89 | Figure 3. Class III Visualization. A three-dimensional representation of predicted cloud structure is shown as translucent, white isosurfaces of cloud water density at 10^-5 kg/kg for August 4, 1996 at 9:00 pm EDT. The cloud surfaces are registered with a terrain map overlaid with coastline (black) and state (white) bo... | [
"Class III enables forecasters to create qualitative three-dimensional representations for both interactive investigation and production of animation via browsing. The consumers may or may not be specialists but the interactive user is likely to be a meteorologist. Thus, the results may be suitable for media and pu... | 3 | Task-specific visualization design: a case study in operational weather forecasting | 1,998 | 10.1109/VISUAL.1998.745330 | http://dx.doi.org/10.1109/VISUAL.1998.745330 | Vis | 100 | null | null | |||
90 | Figure 4. Class IV Visualization. Observations for April 27, 1998 at 8 am CDT are shown. A surface variable (visibility) has been selected for display as pseudo-color, which is overlaid on a topographic map. Rivers (blue) and coastlines (black) are draped on the surface. An upper air variable (relative humidity) is rep... | [
"Class IV provides analysis, viewing, interrogation and interaction tools with standard products designed for AWIPS. This class is similar to the visualization tasks addressed by the aforementioned Vis-5D and RASSIN packages, but with greater emphasis on direct manipulation and the introduction of new realization m... | 4 | Task-specific visualization design: a case study in operational weather forecasting | 1,998 | 10.1109/VISUAL.1998.745330 | http://dx.doi.org/10.1109/VISUAL.1998.745330 | Vis | 15 | null | null |
End of preview. Expand in Data Studio
Dataset Card for SciVisCap
Dataset Summary
SciVisCap is a dataset for captioning scientific visualization (SciVis) figures -- generating a caption for a SciVis rendering, optionally using the paragraph(s) in the source paper that reference it. It contains 3,539 SciVis figures from 992 IEEE Vis and SciVis papers, drawn from the VIS30K corpus and paired with their published captions and figure-referencing paragraphs.
Supported Tasks
- Image captioning (
image-to-text): generate a caption from the rendering alone. - Context-conditioned captioning: generate a caption from the rendering plus its figure-referencing paragraph(s), or from the paragraph(s) alone (a text-only control condition).
Languages
English (en) -- all captions and referencing text are in English.
Dataset Structure
Data Instances
Each item is one figure from one paper: the rendering image, its published caption, the paragraph(s) that reference it, and paper/VIS30K metadata.
Data Fields
| field | type | description |
|---|---|---|
id |
int | unique item id |
image |
image | the SciVis rendering |
caption |
string | published caption, recovered from the VIS30K caption crop via OCR |
referencing_paragraphs |
list[string] | paragraph(s) from the paper body that reference this figure |
figure_number |
int | figure number in the source paper |
title, year, doi, paper_url |
string | source-paper metadata |
conference |
string | Vis or SciVis |
vis_type, encoding_type |
string | VIS30K classification tags |
dim_type |
string | 2D/3D tag (VIS30K classification) |
vis_url |
string | original VIS30K URL for this figure |
cap_url |
string | original VIS30K caption-crop URL |
Data Splits
Splits are made at the paper level (by DOI, 80/10/10). Distributed as three
folders (train/, validation/, test/).
| split | figures | papers |
|---|---|---|
| train | 2,790 | 794 |
| validation | 367 | 99 |
| test | 382 | 99 |
| total | 3,539 | 992 |
Repository Structure
SciVisCap/
βββ README.md
βββ train/
β βββ metadata.jsonl (2,790 items)
β βββ 1996/
β β βββ VisC.73.9.png
β β βββ ...
β βββ 1997/
β βββ ...
β βββ 2020/
βββ validation/
β βββ metadata.jsonl (367 items)
β βββ 1996/
β βββ ...
β βββ 2020/
βββ test/
βββ metadata.jsonl (382 items)
βββ 1996/
βββ ...
βββ 2020/
Images are grouped by split, then by publication year, since VIS30K reuses
figure filenames across different papers/years -- collisions only ever happen
across years, never within the same year, so {year}/{filename} is always
unique.
Dataset Creation
Source Data
Built from VIS30K, a corpus of
figures and tables from IEEE Visualization publications. Items are
restricted to the Vis/SciVis conferences and filtered by vis_type/
encoding_type tags (excluding schematic diagrams, GUI screenshots, bar
charts, and unclassified figures), then manually classified as SciVis,
InfoVis, both, or none by three annotators; only SciVis/both figures are
kept.
Annotations
Published captions are recovered from VIS30K's caption-crop images via OCR. Figure-referencing paragraphs are extracted from the source PDF via Docling and matched to the target figure by number. Recurring OCR errors (e.g. misread figure prefixes) were corrected and re-run; remaining problem cases were manually reviewed and either fixed or excluded.
Considerations for Using the Data
- Figure images and published captions originate from copyrighted papers (Β© their original authors / IEEE); this dataset redistributes them for non-commercial research use only, under the license below.
- A nonzero fraction of candidate figures have no figure-referencing paragraphs at all -- these are excluded from the final dataset, not present with an empty field.
- OCR- and regex-based extraction may leave residual errors despite automated correction and manual review.
Licensing Information
The SciVisCap annotations, extracted figure-referencing paragraphs, captions, metadata, and splits are released under CC BY-NC-SA 4.0. Figure images and published captions remain Β© their original authors / IEEE and are redistributed for non-commercial research use only. SciVisCap is derived from VIS30K; use of the figures is also subject to VIS30K's terms.
- Downloads last month
- 6