Graphics and visualization - the essential features for the classification of systems, J.L. Encarnacao et al ray tracing of particle systems, M. Zeiller simulated multiple scattering for cloud rendering, C. Patmore hierarchical and adaptive meshing with linear interpolation of vertex radiosities, E.P. Lafortune and Y.D. Willems developments in multimedia technologies and learning environments in the UK, R.A. Earnshaw ACC - lossless data compression of animation sequences, E. Groeller and W. Stocker a conceptual model for interaction in multiple representational spaces, K. Seetharaman et al graphics, geometry and mathematical morphology, P.K. Ghosh the slicing extent technique for ray tracing - isolating sparse and dense areas, S.K. Semwal et al the mathematical framework of adjoint equations for illumination computation, S.N. Pattanaik hermite approximation for offset curve computation, V. Ostromoukhov user interface for fashion design, H.M. Werner et al AUTOLAY - an interactive graphics system for the design of aircraft composite components, B.G. Prakash et al an optimal algorithm for computing a best cut of a set of hyperrectangles, F. d'Amore et al three dimensional line clipping by systematic enumeration, N.C. Sharma and S. Monohar a two stage mapping technique for interactive domain discretization, M.A. Dharap and G.R. Shevare the ray casting engine and ray representations for solid modeling - a research synopsis, J.P. Menon realizability in computer aided design, S.G. Dhande and K.P. Karunakaran conceptual surface modeling for industrial design, C.G.C. van Dijk compact representations for planar sections of parametric surfaces, S. Gopalsamy and T.S. Reddy Vinyas - an interactive calligraphic type design system, L. Parida volume modeling for orthopedic surgery, M.S. Kankanhalli et al volume rendering of tetrahedral data, H. Bienenstein et al telecommunications network planning, J.C. Martin multi-dimensional hermite interpolation and approximation for modeling and visualization, C.L. Bajaj. (Part Contents).
An attempt at constructing a realistic force model which is supplied as external force to the dynamic structural model describing the motion of a two-dimensional piece of cloth in free space is described. The model uses a simple and a popular method in low speed aerodynamic theory to represent the distributed force loading on the cloth immersed in an air flow. The model is applied to selected practical problems involving cloth motion. The displacements of various points of the cloth subject to these forces are obtained by numerically solving the cloth deformation model based on the classical theory of elasticity and rigid body mechanics. The resulting motion of the cloth is animated based on the time histories of the numerically computed displacements of the various points on the cloth.
INTRODUCTION: DESIGN METHOD Our work focuses on the design of visualization and interaction techniques for large information spaces. Our design approach is based on a conceptual study of users and their tasks [4]. To face the diversity of users and tasks we adopt an interaction-centered approach and focus on navigation tasks. Based on the characteristics of navigation, we develop ergonomic criteria that need to be satisfied by the designed navigational techniques. One of the criteria, representation multiplicity, is crucial in order to accommodate the diversity of needs. We illustrate representation multiplicity with our own system, VITESSE [7, 10] that runs on the WWW. In the design method, the last step before coding, is software architecture modeling. In the context of VITESSE, we demonstrate the suitability of our architectural model PAC-AMODEUS for representation multiplicity.
This paper describes a method of realistic rendering of a synthetic animated face. Realism is obtained by using texture-mapping which allows rendering surface details, skin grains and other features of the face. Manipulating a 3D model of the face in conjunction with texture-mapping gives complete realistic facial animation. Importance of colors for expressive episode of emotions has also been introduced.
Several attempts aiming at a reduction of the computing time of image sequences by taking advantage of temporal coherence, have been suggested in the literature. Unlike these methods, the approach presented in this paper is independent of the rendering technique. It is mainly based on signal processing, rather than on data structure manipulations. The principle consists in undersampling in time and reconstructing the missing images by motion-compensated interpolation. Although this method does not allow a perfect reconstruction of the missing images, the defects are not perceptible when the sequence is displayed in real time.
Since the last few years, we have been developing the Slicing Extent Technique (SET) for ray tracing. The main difference, between SET and other existing space subdivision techniques, is that 3D-voxels are not used for space partitioning. Instead SET considers the two dimensional projections of the extent surrounding the objects.In this paper, we focus on describing the improvements in the performance, as a result of identifying the dense and sparse areas of the scene. During ray tracing the sparse areas are quickly bypassed. To further reduce the image generation time, SET is applied one more time inside the dense area of the scene.We first describe the Slicing Extent Technique, A uniform partitioning to SET data structure is added to obtain the Modified Slicing Extent Technique (MSET). Next an oct tree is used, during preprocessing, to isolate the blank and dense areas in the scene. However, this oct tree is never used during ray tracing. An analytic comparison of SET and other existing techniques is provided. We also identify future research directions.
This paper describes a method for integrating a particle system into photorealistic images as they can be created with ray tracing. A particle system can be positioned among other objects and thus complex light interactions can be achieved. Particles are arranged in a nonuniformly subdivided bounding box whose voxels contain less than a maximum number of particles. A ray hitting the particle system has to be tested for intersection only with those particles which are contained in voxels which are pierced by the ray, thus eliminating most particles from consideration. During the generation and animation of the particles, they can interact with objects in the environment and they bounce off objects of arbitrary shape. Collision detection is performed using a ray casting technique. The particle system is integrated into a CSG based ray tracing environment and it can be combined with other objects using Boolean operations.
Planning for growth in a telecommunications network is a difficult and labor intensive process. In the current planning process in NYNEX, network facilities are typically represented in large relational databases. There are few tools that allow planners to visualize tile current or future state of the network represented in these databases. The data has many interesting properties: it has a temporal component, it has a geographic component, and there is rich meaning embedded in the alphanumeric strings and in the combinations of records. The volume of data, especially for metropolitan areas, is enormous. Browsing the data currently requires one to write and execute database queries, and then manually analyze the results. This work focuses on tools that allow users to quickly create various types of static and animated visualizations of network data. The current implementation uses a simple but powerful programming language to describe the data and how it maps to the desired visual representation. Using a six degree of freedom input device and a keyboard, the user can navigate through time and space in the database. The system described is implemented on a Silicon Graphics IRIS workstation.
The radiosity algorithm used in image synthesis is regarded as an ideal tool to handle diffuse reflections between surfaces in a global illumination environment. In this paper, a window projection method which extends the conventional radiosity algorithm to deal with both specular inter-reflection and refracted light transmission is described. The method is based on the concept of the ''virtual world'' [9], in which each planar specular, transparent or translucent surface is considered just as a perspective window through which the complex environment can be treated as a pure diffuse environment using the ordinary radiosity solution. Each perspective window is taken both as a medium for calculating light transmission and as a clipping window for final image rendering. By recursively performing the window projection method and employing a kind of random jittering technique, multiple specular reflections and non-ideal specular surfaces can be treated.
There are two basic strategies used for carrying out the illumination computation - the gathering strategy in which light reaching a point from all directions is simulated and the shooting strategy in which light emitted from a point in all directions is simulated. Based on the strategy used, all the existing methods can be classified into two broad categories, namely gathering methods and shooting methods. The radiance equation provides the mathematical basis for the gathering methods and the potential equation provides the mathematical basis for the shooting methods. They together form an adjoint system of equations. In this paper, using the mathematical framework of the adjoint equations we review illumination computation methods, categorising them as using the gathering or shooting strategy or both. Another basis for categorisation is the basic equation solution strategy used, namely deterministic or nondeterministic.
In this paper we address the problem of representing trim curves corresponding to planar sections of parametric surfaces. We represent a trim curve as a composite Bezier cubic or quintic curve in the parameter space of the surface with geometric continuity of order 1 (G(1)) or 2 (G(2)) respectively, using an error minimisation technique to fit optimal Bezier cubics or quintics between trim curve nodes.
In this paper we use some well known theorems of algebraic geometry in reducing polynomial Hermite interpolation and approximation in any dimension to the solution of linear systems. We present a mix of symbolic and numerical algorithms for low degree curve ts through points in the plane, surface ts through points and curves in space, and in general, hypersuface ts through points, curves, surfaces, and sub-varieties in n dimensional space. These interpolatory and (or) approximatory ts may also be made to match derivative information along all the sub varieties. Such multi-dimensional hypersurface interpolation and approximation provides mathematical models for scattered data sampled in three or higher dimensions and can be used to compute volumes, gradients, or more uniform samples for easy and realistic visualization.
Mathematical representation, graphics rendering and physical realization of a shape constitute three important aspects of any computer-based design and manufacturing process. Parametric representation of shapes has been found to be suitable from graphics rendering as well as engineering analysis considerations. However, from the view point of design for manufacturing a shape, it has been found that additional geometric considerations need to be taken into account. The present paper outlines two such models, viz., the conjugate geometry and the intrinsic geometry models. It has been shown that conjugate geometry can be used to model several different machining processes. The intrinsic geometry model seems to hold considerable potential to model metal forming processes. Mathematical details of conjugate geometry and intrinsic geometry have been discussed with the help of a few illustrative examples.
Realistic rendering of clouds remains a stumbling block for computer graphics. Previous work has produced excellent images for certain cloud types, typically where there is little variation of brightness over the image of the cloud and so a single scattering solution is sufficient. Such models are not sufficient for clouds with high albedo. For these clouds we need to consider complex scattering phenomena to approximate the radiation of light through the volume. We formulate a local solution of the transfer equation for a discrete directional model and present a method of simulating multiple scattering on a cubic lattice by progressively applying a local solution. Individual views of the solution are generated by volume tracing. An octree structure is used to hold radiant energy information. Incident radiance can be from any source, such as the sun, the sky or the ground and emitted radiances can provide both illumination and shadowing information.
This paper discusses different interpolation schemes for the radiosity method. It is shown that similar mathematical expressions exist for piece-wise constant approximation and linear approximation of the radiosity function. The latter improves accuracy and does not necessarily entail a higher complexity. Our implementation of a linear interpolation scheme demonstrates that hierarchical and adaptive meshing can be used elegantly.
The terms volume rendering and volume visualization refer to techniques for displaying scalar fields of three spatial dimensions known as volume data. A general model to generate images from such data is to simulate the interaction of light with a semitransparent material based on linear transport theory. There is a wide variety of possible mappings of function values onto the ''physical'' model parameters. Starting from this basis we develop a very descriptive approach for volume visualization and solve it for the case of linearly changing data. Most methods assume function values given on the nodes of a regular cartesian grid. However, sample points are often provided on irregular scattered locations. An efficient technique is described for directly ray-casting volume data defined as tetrahedral cells. It can also be used for interpolating such data to a regular grid.
Photorealistic rendering techniques, such as raytracing, have very high computational demands. Parallel processing may be used to reduce these computation times. However, complex scenes may contain far more data that can be accommodated at each processing element in a multiprocessor system and so effective data management strategies are necessary to avoid significant delays while data is fetched from remote locations. This paper proposes a data management strategy which has more than one application process at each processing element in combination with a global broadcast mechanism and message reduction techniques. This strategy will allow complex images to be raytraced using parallel processing in reasonable times.
Representation of freeform surfaces has always been a problem for industrial designers. CAD systems represent surfaces adequately, but do not support the special requirements of the conceptual design phase. We have developed the 'Fast Shape Designer' conceptual surface modeler based on these special requirements. The modeler supports surfaces defined over a network of hand-sketched curves; correction of any curve immediately causes recalculation of surfaces. We describe this modeler and several novel features. We show how the flexible object-oriented data structure captures topological information, and how it is tied to the graphical user interface. We also present our surface construction algorithm which interpolates transfinite data, and we propose an improvement upon an extension oi the Gregory patch.
Advances in computer graphics in the recent twenty years have stimulated different schemes to classify research directions and systems. In the early days, graphics systems were identified to be vector or raster graphics in terms of technology. Sutherland`s Sketchpad system was the first example that allowed to distinguish between passive and interactive computer graphics. Dimensions of the geometric data model classified systems to be a 2D or 3D system. This scheme was used by standardization activities in computer graphics during the last decade. However, the approaches of standard committees to develop a reference model for computer graphics have shown very clearly that the variety of systems and the complexity within one graphics system prevents from the establishment of an easy-to-understand model. Taxonomies in scientific visualization (MzCo-87) focussed on the integration of different disciplines like computer graphics and computer vision, and the use of available, mostly heterog eneous system components and peripherals. At least, scientific visualization has shown very clearly that computer graphics today is very different from drawing and image processing. Our understanding (Felg-90) of scientific visualization comprises outstanding system requirements like: massive amounts of complex and multidimensional data to be processed, peripheral and algorithmic means for interactive data exploration, manifold alternative (physical and logical) visual (but also non-visual) data presentation techniques, and computational models for physical phenomena. Consequently, scientific visualization requires a correspondence between the human perception and the abstract computer-internal representation of the physical world. Visualization in scientific computing needs this correspondence, virtual reality even requires more! Virtual reality presumes integrated presentation, feedback and simulation techniques and demands realtime! Realtime in this context is defined as the evaluation of the computational model to present continuity for the human perception. Realtime is obtained by an image refresh rate of at least 10 frames per second for the visual senses and by an 8 kHz sample rate for the ausitive senses. (IGD)
