Skip to main content



Interdisciplinary Techniques, Toolkits and Models for Scientific Visualization

Scientific visualization is concerned with the processing and display of large amounts of data in such a way as to facilitate assimilation of the information in an efficient and effective manner. Many disciplines and application have common requirements. This paper summarises current work in this area in the UK and the range of tools and techniques currently available.
A Scientific Visualization Workshop in the UK in 1991 examined the areas of framework models, visualization techniques, data facilities, human-computer interface aspects, applications, and products. Studies were also done in the area of enabling technologies — hardware, software, toolkits, database systems, and generic data formats. This work has been published as ‘Scientific Visualization — Techniques and Applications’, Springer-Verlag, 1992. This paper summarises the key results from this work.
An inter-disciplinary Scientific Visualization Group was formed at the University of Leeds in October 1990 to facilitate and coordinate campus-wide developments in area of tools and methodologies for visualization. We summarise the current work and objectives of the group in interdisciplinary application areas, and also current studies in methodologies for visualization via networks. Experiments are being set in place to study the partitioning of the visualization pipeline across different facilities (e.g. parallel processor, supercomputer, graphics workstation) and investigate interactive response times and network bandwidth requirements. Optimum configurations should enable local users to utilize high-powered facilities via network access. Desk top visualization is part of the required user toolset for the 1990’s.
R. A. Earnshaw

Strategies and Implementation of Ray-Tracing with Increasing Realism

Images produced by the ray-tracing algorithm are considered as having a good quality and a high level of realism. Still the algorithm can not be used in low price systems due to the great demanding of processing time required.
To help solving this problem we have been enhancing a conventional raytracing program in order to have less quality images in a shorter time. Such images can be continuously improved without loss of previous results and with a low time overhead if compared to a traditional ray-tracing. In this work, problems related to first images production (few rays to give the best quality) and images updating (minimum repetition of processing) are mentioned and efficient solutions are presented.
Strategies to produce increasing realism images by the manipulation of the sample rate and the maximum depth level are presented.
J. M. Leitāo, A. A. Sousa, A. C. Costa, F. N. Ferreira

Directional Search Algorithms for Visualizing and Mixing Star-Shaped Grids

This paper describes different algorithmic approaches to solve some basic problems encountered in the building of various visualization tools intended for the grids generated by scientists. The visualization of the scalar and vector fields in such grids can lead to “traditional” geometric problems: point inclusion, point location, line intersection and line location. To be computational fast, any algorithm to solve these problems must appropriately exploit the geometric or even “topological” properties of the grids studied. We have noticed the frequency of a special type of grid that is called here “star-shaped”. We will show how the interesting nature of such a grid, concave but “not too concave”, can be used to build efficient algorithms for solving these problems. These algorithms are based on the adjacency properties of the grids used and on the simple remark that such problems can be reduced to 2D and ID problems by restricting the search to appropriate planes and lines. An extension to general grids is suggested, which takes full advantage of the special-purpose processors of modern graphics workstations. Some examples of applications are stressed, in particular for “slicing” grids with other grids and mixing grids together.
J. L. Pajon

The Cone Of Vision: A new Technique for Interactive Scientific Visualization

A new algorithm is proposed for interactive visualization of large 3D timedependent datasets. The idea is to process only the part of the data which is actually seen by the observer’s eye i.e. the Cone Of Vision (COV). This technique improves interactivity and limits memory requirements since only the COV is stored in the core memory. This feature is very useful for very large and/or timedependent datasets. The COV approach has been implemented in a scientific visualization software called ICFS and it appeared to be up to 100 times better than a classical approach.
D. Astruc, A. Vincent

Real-Time Visualization of Two-Dimensional Fluid Flow

The spread of personal computers with high-quality graphics displays and increasing of computational speed, is bringing graphics to more users and allows a wide range of applications. The main purpose of this paper is to give a physical-intuitive understanding of fluid dynamics to engineers and students by the presentation of fluid flow examples.
The work presented here describes the physical fundamentals and the numerical algorithm used to perform calculations and visualization of incompressible two-dimensional flows. The model has been applied to a wide variety of flow calculations and examples are shown in the paper emphasising the versatility that can be obtained.
J. J. D. Domingos, M. P. N. Aguas

Visualization of Flow Simulation Data in Environmental Modeling

This paper introduces a flow visualization system and an application in environmental modeling. A collection of 2D and 3D flow visualization methods suitable for flow simulation data are discussed. Emphasis is laid on techniques related to the simulation of the propagation of massless particles in the flow, like particle tracing and particle pathways. The scientific visualization system apE together with a set of enhancements for flow visualization forms a generic flow visualization system. The resulting system is used for evaluating simulation data provided by numerical models of water flows.
V. Jung

Visualization of Volumetric Flow-Field Data

The present paper is concerned with two issues of scientific visualization, fast interactive slicing and volume rendering of three-dimensional fluid flow data containing complicated three-dimensional flow structures and possible errors. Both issues are included in a single program called Voxvu, which is traded by Sun Microsystems, Inc. together with its TAAC-1 application accelerator hardware [TAA89]. Although the discussion relies on a program running only on a single hardware platform, some general conclusions can be drawn. From the experience gained in using this program, we may conclude that fast interactive visualization is the most effective, and therefore most powerful tool for the interpretation of three-dimensional fluid flow data.
Volume rendering [DRE88] was found to be an effective and easy-to-use tool for the visualization of flow structures. During further experimentation with the method, the question arose as to what significance should be attributed to different substances in a single scalar field. A method was finally found where different substance properties are attributed to different vector components of a vector field, thus creating a new and meaningful rendering of fluid flow data. Discussed first is the problem that needs extensive use of visualization. The simulation and hardware environment are then described followed by the discussion of the visualization method. Specific results and experience are also reported.
U. Rist

Multivariate Modeling of the Dynamics of Charged Particles

The trajectories of charged particles in a constant magnetic field are circles with radius dependent on the strength of this field. The presence of a travelling wave normal to this field perturbs these trajectories. When the interaction parameter exceeds a certain limit, the resultant motion becomes chaotic, dividing the phase space into stable regions, separated by stochastic boundaries. Under certain conditions, the particle can be accelerated to very high energies. Pictorial analysis provides the insight to the complexity of modes that are present and delineates the range of parameters best suited for efficient acceleration. Data generation is done using the Bulirsch-Stoer integrator that dramatically reduces processing demands. The analysis is made by building a 3D image using data slices. Subsequent planar cross sections as well as solid cutouts yield details about interconnection amongst various parameters in this interacting system.
M. M. Novak

Algorithm Animation of Computational Chains

A_VISTA (Visualization & Interactive Steering for Task Activation) is part of the modular computer-aided control engineering environment ANDECS1 and provides a framework for the visualization of engineering computations. Available graphical modules can be used both for offline & online visualization, interactive steering, and information zooming. Multiple graphical displays in multiple windows are supported. A_VISTA makes full use of databased integration of software modules.
A graphical modeling set up environment for macros is provided to facilitate the specification of complex computational and visualization tasks, e.g parameter studies via interactive steering. The development of both A_VISTA and the macro-editor is along the lines of the research area of Algorithm Animation. Since the modules of AVISTA are not bound to data produced by a special algorithm, they can be used for actively exploring and monitoring the dynamic behavior of any computational procedure. This makes it well suited for the algorithm animation of general computational chains.
R. Finsterwalder

Towards an Object-Oriented Kernel for Geometric Modeling

The storage and management of relationships among elementary geometric objects, other than the topological relationships, is desirable in Geometric Modeling. Among them, the relationships derived from the construction of geometric objects and from their relative positions may have special importance, since they can influence the next work steps.
The main goal of this paper is to present a kernel for geometric modeling based on object-oriented programming concepts and able to support, in a unified way, different kinds of geometric objects and relationships between them. Using the object-oriented paradigm it is possible to regard relationships as a class of objects and, therefore, work at the same level with geometric objects and the logical relationships among them. Test and satisfaction functions for the relationships (constraints) are included in a methods library, and are used to implement constraint solvers.
J. C. Teixeira, V. Sakas

A Cellular Approach for Feature-Based Modeling

The main goal of this paper is to present a new cellular scheme for feature-based modeling and to show its applicability. With this cellular scheme it becomes possible to distinguish, clearly, and completely, between the morphology and the geometry of a solid. An object is globally and morphologically defined by a collection of interacting form features. Moreover, a form feature is defined as a structured cluster of one or more volumetric cells. As it will be shown, this conceptual scheme facilitates reasoning strategies, at the cellular level, fired from constructive and destructive morphological actions taken at the feature level.
A. Gomes, R. Bidarra, J. Teixeira

Issues on Feature-Based Design Using a Hierarchical Data Scheme

Computer-aided design based on the notion of features is currently being discussed as a means for better capturing the involved semantics, i.e. the designer’s intent. This paper proposes a hierarchical data scheme for feature-based design with particular emphasis on early phases of design, such as functional and conceptual design. The top level reflects the designer’s view on the function of a product part and consists of specifications of functional features. At the second level, features are usually defined as generic shape characteristics (distinguishable portions of a shape) which have no application-specific meaning and no presumptions to the kind of representation. The third level is concerned with the internal representation of features in different ways and their geometric evaluation using conventional geometric modeling techniques. The data relationships occurring in the scheme are performed by mappings in top-down and bottom-up fashion.
J. Ovtcharova

Product Modeling: An Approach for Integrated Information Processing

Integrated information models build the basis for product modeling. The paper describes a general framework of integrated models with special focus on product models. Then different approaches to product modeling are discussed. Furthermore two sample product modeling tools are introduced. A feature modeller supports shape oriented manipulations on the product, a product structure editor enables the designer to handle structural information.
F.-L. Krause

Direct Graphic User Interaction with Modelers Based on Constructive Solid Geometry

Boolean operations provide convenient methods for defining solid objects, and are supported by most of the solid modelers available today. Graphic User Interfaces (GUIs) for solid modelers typically are based on picking edges and vertices on a display, and correlating them with entities in the boundary representation (Brep) of a solid. The graphic operations involved are fast, but they require that a Brep be computed, and this is computationally expensive. This paper discusses an alternative approach, which does not require the existence of a Brep, and is based on correlating pick operations directly with the surfaces of Constructive Solid Geometry (CSG) primitives. The new approach, in conjunction with recently-developed, special-purpose, modeling hardware, should lead to higher performance user interfaces than those currently available. The initial implementation of a GUI based on these concepts is presented. The interface uses the X/Motif tool kit, and interacts with the PADL-2 CSG modeler through Unix interprocess communication techniques.
L. M. Encarnaçāo, A. G. A. Requicha

Constructing Free-Form Surfaces for Milling from Digitized Data

Using a three-axis miller, copies, negative moulds, etc. of a solid body shall be made. To reduce machine time, a method is presented so that the body has to be scanned only once, even if differently sized tools are used. This is accomplished by approximating the given surface with B-splines and generating the milling paths from this mathematical description. Prior to being fed into a milling machine, these control programs and several other data can also be displayed on a screen, in order to judge the quality and smoothness of the result. A small example is included.
F. Anheuser

Visual Simulation of the Geodesic and Non-Geodesic Trajectories of the Filament Winding

To improve the efficiency of the winding process its simulation is necessary. Through the simulation, not only the most appropriate filament winding for a specific purpose can be chosen, but also wrong winding configurations can be avoided as well. An interactive environment for the analysis of the filament winding process has been developed and is described in this paper. Suitable algorithms for computing geodesic and non-geodesic trajectories are presented.
J. A. O. Simões, Shin-Ting Wu, F. Loseries

Cooperative Street-Lighting Simulation by Multi-Disciplinary Users

Two of the recent areas of research in Computer Graphics are street-lighting simulation and cooperative work. In the first, the user can generate images that give information about the perceptibility of streets at night. The second deals with specific difficulties posed when a certain problem is tackled by several users simultaneously. To work in the first area the users need to have proficiency at different levels, namely at the designer level, at the technical and at the legal level. As it is difficult for any user to be expert in all of these, problems can arise. We propose computer supported cooperation of multi-disciplinary users as a way to solve some of these problems. In this paper we describe an extension to LISA (Street- Lighting Simulation System) to support cooperation of multi-disciplinary users. The extension is made in three directions. The first is to add a cooperative user interface to enable a group of users to build scene descriptions cooperatively. The second is to enable different users to visualize separately the rendered results of the simulation. Finally the third direction, very important to enhance the group effectiveness, is to support communication among users.
A. Santos, S. Clavé

A CAD-Based System for Air-Conditioning in Buildings

Existing simulation tools related with energy analysis in buildings are not flexible enough and the amount of training time required to reach high proficiency is beyond the resources of most design studios.
This paper describes the present on-going development of a general energy simulation tool called GraphBase. The priority target is the automation of input data addressed to energy efficiency and associated analysis in buildings. Therefore, a more user-friendly “front-end” is now being proposed and presented using a hierarchical database structure through a commercially available CAD environment.
By means of a brief description of its architecture and concepts, some aspects of an application in a real building performance evaluation context are shown.
J. M. S. Dionisio, J. J. D. Domingos, H. Roth

Language Support for Raster Image Manipulation in Databases

Multidimensional array data come up in many application areas. In computer graphics and imaging, those — usually 2D — arrays are conceived as raster images; pixel information, then, denotes some color value. In scientific visualization, pixel or voxel information can carry arbitrary semantics, such as temperature, speed, or stress. The size of such structures may well go into Gigabytes per object.
In principle, storage of huge data volumes and flexible retrieval among them is a typical task of database systems. However, current database technology is not prepared to cope with multidimensional arrays of arbitrary size. Hence, if today in visualization database systems are employed at all, they store such data as byte sequences, thereby losing all structure information. As a consequence, it is impossible to extract partial information from one such object, or to use it within a query. Moreover, as structure information is lost, transparent exchange within heterogeneous networks cannot be supported.
This paper describes an approach to the modeling of general arrays of unlimited size in database systems. It is done in a way that the system keeps structure information in the schema, and hence overcomes the previously stated limitations.
As the most prominent case of multidimensional arrays still are raster images, focus here is on the two-dimensional case. The results, however, are valid for any number of dimension.
P. Baumann

DEDICATED-Development and Installation of a Common Architecture for Europe-Wide Computer-Based Education and Training

The project DEDICATED [DEDI92] is the answer of the DEDICATED consortium1 to the European needs in the areas education, teaching, and training and to the world-wide rapid advances in information technology. The results may play an important role in the European integration process, which does its next step with the begin of the next year.
R. Lindner


Weitere Informationen