Barrera,T., Hast,A., Bengtsson,E.
Abstract:
Quadratic shading has been proposed as a technique giving better results than Gouraud shading, but which is substantially faster than Phong shading. Several techniques for fitting a second order surface to six points have been proposed. We show in this paper how an approximation of the mid-edge samples can be done in a very efficient way. An approximation of the mid edge vectors are derived. Several advantages are apparent when these vectors are put into the original formulation. First of all it will only depend on the vertex vectors. Moreover, it will simplify the setup and no extra square roots are necessary for normalizing the mid-edge vectors. The setup will be about three times faster than previous approaches. This makes quadratic shading very fast for interpolation of both diffuse and specular light, which will make it suitable for near Phong quality software renderings.
Madi,M.
Abstract:
The need to systematically generate sets of reference points with prescribed arclengths along parametric curves, with accuracy and real-time performance usually arises in applications related to CNC machining, highway and railway design, manufacturing industry, and animation.
Mechanisms to produce a parameter set that yield the coordinates of prescribed reference points along the curve ${\bf Q}(t) = \{x(t), y(t)\}$ are therefore sought.
Arclength parameterizable expressions usually yield the parameter set that is necessary to generate the reference points; however, for typical design curves, such expressions are often not available in closed form. It is desirable to find efficient ways to compensate for lack of arclength parameterization.
In this paper, several methods for approximating arclength parameterization are studied. These methods are examined for both accuracy and real-time processing requirements. The paper also introduces a numerical interpolation technique for a cubic interpolator function; the interpolator exploits the influence of end point tangent vectors to generate approximately uniformly distributed reference points as an example application.
Santonja,J., Linares,J., Cuesta,D., Mico,P.
Abstract:
In this work a new image deformation method is described based on a copy-art technique. This copy-art process involves the use of either a photocopier or a linear scanner with a motif in motion as the light bar sweeps by to create distortions. The new method provides an easy-to-use tool which is very intuitive and offers predictable results. This allows the artist to recreate the original copy-art process digitally. In order to define the transformation completely, the user has to specify three elements for the original image interactively: path, rotation and velocity. To get the final image, two methods can be followed: either to sample each line from the original image to the result image using forward mapping or to split the transformation in a grid. This latter approach will allow more control in the final result.
Mousa,M., Chaine,R., Akkouche,S.
Abstract:
3D meshes are the most common representation of 3D models. However, surfaces represented by 3D meshes may contain noise or some unrequired details. Multiresolution representations and filtering techniques are very useful in this case. In this paper, we propose a new and compact representation for the surface of a general 3D mesh using the spherical harmonics. This representation can be useful in many applications like filtering , progressive transmission and compression of 3D surfaces. A first basic framework is designed for star-shaped objects. Then, we show how this framework can be extended easily to general form meshes using certain segmentation techniques in combination with implicit surface techniques. Another interesting point of our approach is that the computation of the involved spherical harmonics transform is decomposed into the computation of spherical harmonics transforms based on elementary triangles that compose the mesh. This makes the complexity of the computation of the used spherical harmonics transform linearly dependant on the number of triangles of the mesh. Some experimental examples demonstrating our method are presented.
Ilmonen,T., Takala,T., Laitinen,J.
Abstract:
Fire, explosions, and other special effects are often created with particle systems. In real-time applications the particle systems must be very fast to evaluate since otherwise the application cannot maintain reasonable frame rate. One part of this challenge is the collision detection between particles and the objects in the scene. We present a new approach to collision detection and surface flow effects for particle systems. In pre-processing phase we rasterize a 3D model into a distance/normal grid. The grid can be used for collision avoidance, to create surface drag and to simulate fluid flow around non-deforming objects. This method is not physically accurate, but it provides visually plausible results. The primary benefit of this method is that it is efficient and its performance is independent of the complexity of the model. This methods works well in real-time, in some cases surpassing the rendering speed of modern graphics hardware by order of magnitude.
Maass,S., Döllner,J.
Abstract:
We present a technique for the dynamic annotation of three-dimensional objects in interactive virtual environments.
Annotations represent textual or symbolic descriptions providing explanatory or thematic information
associated with scene objects. In contrast to techniques that treat annotations as two-dimensional view-plane
elements, our technique models annotations as separate three-dimensional scene elements that are automatically
positioned and oriented according to the shape of the referenced object. The shape of such an object is generalized
by an annotation hull and skeleton used to determine an adequate position and orientation of the annotation
with respect to the viewing direction. During camera movements, annotations float along the surface of the annotation
hull. Additional constraints for the generalizations provide further control about geometric and dynamical
properties. In a case study, we show how this technique can be applied for annotating buildings and other
components of virtual 3D city models.
Lam,R., Rodrigues,J., du Buf,J.
Abstract:
In this paper we present a brief overview of the processing in the primary visual cortex, the multi-scale line/edge and keypoint representations, and a model of brightness perception. This model, which is being extended from 1D to 2D, is based on a symbolic line and edge interpretation: lines are represented by scaled Gaussians and edges by scaled, Gaussian-windowed error functions. We show that this model, in combination with standard techniques from graphics, provides a very fertile basis for non-photorealistic image rendering.
Vichitvejpaisal,P., Kanongchaiyos,P.
Abstract:
A set of billboards can represent 3D models for extreme simplification in real-time rendering. Unlike conventional polygon method, billboard-based technique has the rendering time of the model proportional to its contribution to the image. Thus, it has an automatically built-in Level of Detail. However, previous techniques still have a limitation of viewing angle and non-accuracy of model representation. We present an adaptive rendering method using enhanced billboards. First, each enhanced billboard, representing portion of the model, is defined to have four maps: depth map, normal map, color map and transparency map. The model is then projected to a number of viewing planes in difference viewing directions. Consequently, these enhanced billboards are rendered based on ray-height-field intersection algorithm implemented on GPU. This representation can maintain geometry and silhouette of the model with no limit viewing direction. Moreover, real-time rendering is supported.
Li,L., Zhao-qi,W., Deng-Ming,Z., Shi-Hong,X.
Abstract:
The existing motion editing methods don't take collision between the limbs into account potentially producing implausible motions. Facing this problem, in this paper we present an integrated framework of motion editing for producing collision-free motion in real-time. We first provide an efficient scheme for collision detection based on skeleton model, which can fast find the penetration between limbs of human body. Then we provide a novel scheme for constraint generation, which embeds the motion characteristics and inter-frame coherency into the constraint generator, and preserves the qualities of the original motion and motion consistency in the target motion. Finally, we show how to apply the Kalman filter to constraint resolve in a real-time manner. Comparing to other existing methods, our method not only can obtain the collision-free motion in real time, but also can preserve the original characteristics as many as possible and generate the new motion similar to the original motion. Experiment shows that our approach is very useful in producing natural and collision-free motion and efficient enough for application in animation system and game.
La Greca,R., Daniel,M.
Abstract:
B-Spline and NURBS surfaces are most often considered to model objects. The object shape is designed by manipulating several control points, which is often very complex and tedious. The declarative approach of surface modelling is a fast and easy way to obtain sketches of parametric surfaces. The designer provides a description of the shape he/she wants to obtain. The semantic extracted from this description is structured through XML language. As a result, a set of parametric surfaces corresponding to the given constraints and features is proposed to the user. This approach is specially devoted to speed up the preliminary design process. This paper introduces our system as a high level tool of surface modelling. Details dealing with the different models and processing involved in our system are proposed. The document is illustrated by the first results of our research study.
Shanbhag,S., Chandran,S.
Abstract:
The notion of transplanting limbs to enhance a motion capture database is appealing and has been recently introduced [Sha04], [Ike04]. A key difficulty in the process is identifying believable combinations. Not all transplantations are successful; we also need to identify appropriate frames in the different clips that are "cut-pasted."; In this paper, we describe motion grafting, a method to synthesize new believable motion using existing motion captured data. In our deterministic scheme designed for locomotive actions, motion grafts increase the number of combinations by mixing independent kinematics chains with a base motion in a given clip.
Our scheme uses a cluster graph data structure to establish correlation among grafts so that the result is believable and synchronized.
Mashtalir,S., Shcherbinin,K., Yegorova,E.
Abstract:
Permanently a special emphasis is given to image processing considering geometric distortions for remote sensing, and image tracking especially. We discuss an approach to affine transformation parameters search on the base of salient points which can be external and internal relative to shapes. To find affinity we use sets of image binary cuts and one-to-one point correspondence. This correspondence is determined by means of the affine-equivalent shape ratio invariance. We investigate two types of salient points on the base of binary morphology. First ones are vertices of convex hulls and second ones belong to skeletons. Finally, solution of overdetermined system of linear equations gives us affine transformation parameters. Results of the exterior points and skeleton normalization methods comparative analysis are discussed.
Kovács,L., Szirányi,T.
Abstract:
We present an automatic two dimensional model based non-photorealistical painterly rendering method which uses automatic relative focus map extraction from the model image to produce a relevance-based multilayer painting. Using relative focus segmentation, the painterly rendered image will be built as a series of differently detailed layers.
Lanquetin,S., Neveu,M.
Abstract:
Reverse subdivision consists in constructing a coarse mesh of a model from a finer mesh of this same model. In this paper, we give formulas for reverse Catmull-Clark subdivision. These formulas allow constructing a coarse mesh for almost all meshes. The condition for being able to apply these formulas is that the mesh to reverse must be generated by subdivision of a coarse mesh. Except for this condition, the mesh can be arbitrary. Vertices can be regular or extraordinary and the mesh can be arbitrary (triangular, quadrilateral…).
Sitte,R., Cai,J.
Abstract:
In this paper we introduce our virtual etching as part of MAGDA a CAD system for Micro Electro Mechanical Systems (MEMS). Virtual prototyping visualizations require fast algorithms for visualization that are suitable for interactive design. Modern MEMS simulators do not offer dynamic visualizations for etching. Etching progress is time dependent, typically calculated with Finite Element Analysis, which is too slow calculating, hence not suitable for interactive design. Etching progress is important in MEMS less than 10mm, where Silicon technology must be used, with its repeated cycles of deposition and lithography/etching until the desired structure is formed. While etching performance is well known from the Integrated Circuit processing, it is not so predictable in MEMS because the shapes are more complex. Underetching is undesired in IC technology, but it is crucial in shaping MEMS structures. We use a Marker/String method for the progressive mesh as a faster method suitable for interactive design. The method is not known much for etching; but used in other applications. We have found a way of overcoming swallowtail conditions that appear on corners. We are also able to simulate underetching. In this paper we demonstrate the progress of etching using a circular lithography mask calculated in 2D then rotated, and a square mask calculated in 3D. In both cases we are able to simulate underetching. The method can be extended into larger material removal CAD visualizations. In this way we made a step towards filling a long existing need in virtual prototyping.
Tse,R., Gold,C.M., Kidner,D.
Abstract:
Estimating building forms from LIDAR data has usually been done by attempting to fit standardized building types to the residual data points after an estimated “bare earth” terrain surface has been removed. We propose an approach based on segmenting the raw data into “high” and “low” regions, and then modelling the walls and roofs by extruding the triangulated terrain surface (TIN) using CAD-type Euler operators. The segmentation may be done by the addition of building boundary data to the TIN so as to force triangle edges to match the boundaries, and then using Euler operators to extrude the building, producing vertical walls rather than the more usual sloping walls formed from TIN models alone. If boundary data is not available then an automated segmentation method based on adaptive Voronoi cells may be used, so that each cell contains either “high” or “low” LIDAR data, but not both. This prismatic model, with flat-topped cells, approximates the building forms without hypothesizing specific building types. Once the segmentation is achieved, and the walls constructed, we attempt to model the roofs by calculating the eigenvalues and eigenvectors of the vector normals of the TIN sections within the building boundaries. The smallest eigenvalue gives the predicted roof orientation, and the resulting roof profile is then modelled.
Roth,M., Riess,P., Reiners,D.
Abstract:
Sort-first or image space parallel rendering has been the subject of a significant amount of research. Its limitation for scalability is well known. Nevertheless, sort-first has some advantages over other approaches: it has moderate and predictable communication overhead, and it has no problems with multipass or transparent rendering. Especially for
multi-projector display systems, driven by a cluster of standard PCs with limited network bandwidth, it is the first choice to achieve interactive frame rates. In this paper we present an approach to do load balancing for arbitrary multi projector systems, based on sort-first.
Arya,A., DiPaola,S.
Abstract:
Modern multimedia presentations are aggregations of objects with different types such as video and audio. Due to the importance of facial actions and expressions in verbal and non-verbal communication, the authors have proposed Face Multimedia Object as a new higher-level media type that encapsulates all the requirements of facial animation for a face-based multimedia presentations within one single object. In this paper, Interactive Face Animation - Comprehensive Environment (iFACE) is described as a general-purpose software framework that implements the Face Multimedia Object and provides the related functionality and tools for a variety of interactive applications such as games and online services. iFACE exposes programming interfaces and provides authoring and scripting tools to design a face object, define its behaviours, and animate it through static or interactive situations. The framework is based on four parameterized spaces of Geometry, Mood, Personality, and Knowledge that together form the appearance and behaviour of the face. iFACE can function as a common face engine for design and run-time environments to simplify the work of content and software developers.
Yan,W., Song,M., Bu,J., Chen,Ch.
Abstract:
Real-time rendering of a complete 3D scene with hatching strokes is an important direction in NPR field. In this paper, a comprehensive solution is presented to render complicate scenes with pen-and-ink style in real time. With the help of powerful programmable graphics hardware, our real-time system includes many features as hatching, continuous tones, silhouettes, and shadows. We build our approach in image-space, while allowing for the stroke directions and frame coherence.
In our method, various features of pen-and-ink drawings are derived from 3-D information through multi-pass rendering. After synthesis, the desired shading tone is achieved by mapping preprocessed stroke textures to the screen. As a tip for saving texture memory, we prepare a few stroke textures with only certain tones and directions, and compose requisite stroke types in real time. Furthermore, we develop some forms of indication to convey the impression of a texture without drawing every single stroke, which makes the result look more natural and art-stylized.
Piranda,B.,Magdelaine,S.,Arques,D.
Abstract:
This paper expounds a new method of complex surfaces rendering permitting to visualize atlas of discoids in real-time. The atlas of discoids allow to define surfaces very easily without the topological constraints we could get with the classical methods such as polygons meshes ones. Our basic model is completed by many algorithms permitting either to reconstruct implicit surfaces or to calculate the global illumination using radiosity algorithm. But up to now, no fast viewing method has been proposed for this modeling algorithm.
We present here a real time rendering algorithm which exploits the recent extensions of OpenGL library and the possibilities offered by GPU like the shaders programs to perform non standard fragments mixing operations.
Ahrenberg,L., Magnor,M.
Abstract:
This paper presents a light field rendering framework based on matrix optics. Matrix optics, in contrast to intersection-based methods such as ray-tracing, has
the advantage that a generic series of optic operators can be combined into a single matrix.
This enables us to realize a "virtual optical bench" where different setups can be easily tested.
We introduce the theoretical foundation of matrix optics and define a set of operators suitable for light fields. We then discuss a wavelet compression scheme for our light field representation. Finally we introduce a real-time rendering approach based on matrix optics suitable for both uncompressed and compressed light fields.
Battiato, S., Di Blasi, G., Farinella, G.M., Gallo, G.
Abstract:
In this paper we present a method to generate a digital mosaic starting from a raster input image. Mosaics generation of artistic quality is challenging. The basic elements, the tiles, typically small polygons, must be packed tightly, emphasizing orientations chosen by the artist. An ad-hoc boundaries detection have to be performed according to the directional guidelines. Different mosaic styles can be automatically rendered, depending on artistic techniques (?opus musivum?, ?opus vermiculatum?, etc.) considered.
The proposed method is able to reproduce the colors of the original image emphasizing relevant boundaries by placing tiles along their direction. The boundaries detection is based on the statistical region merging algorithm. In particular the technique is able to reproduce the ?opus vermiculatum? mosaic style.
Several examples reported in the paper show how the right mixture of mathematical tools together with century proved ideas from mosaicists may lead to impressive results.
Beccari, C., Casciola, G., Romani, L.
Abstract:
In this paper we present a novel subdivision scheme that can produce a nice-looking interpolation of the control points of the initial polyline, giving the possibility of adjusting the local shape of the limit curve by choosing a set of tension parameters associated with the polyline edges. If compared with the other existing methods, the proposed model is the only one that allows to exactly reproduce conic section arcs of arbitrary length, create a variety of shape effects like bumps and flat edges, and mix them in the same curve in an unrestricted way. While this is impossible using existing 4-point interpolatory schemes, it can be easily done here, since the proposed subdivision scheme is non-stationary and non-uniform at the same time.
Benes,B., Dorjgotov,E., Arns,L., Bertoline,G.
Abstract:
We present a novel approach for a virtual granular material interactive manipulation with force feedback. A user can interactively change a height-field model of sand by dragging objects inside. The virtual sand behaves like real sand moving to the sides and falling back, filling holes and irregularities on the surface. The dragging object position is controlled by a haptic device that provides a position in 3D space that correspondingly changes the sand model. The sand model provides force feedback to the haptic device resulting in two principal forces; the repulse that has vertical direction and the viscous drag. The resulting force is delivered back to the haptic device. The user can sense the sand's response as the dragging object moves through the virtual sand. This results in the haptic-visual feedback providing a higher degree of plausibility than visual feedback alone.
Dexet,M., Andres,E.
Abstract:
In this paper, we present the design of a topological based geometrical modeler kernel. With this modeler, our goal is to create, import and handle geometrical objects represented both in a rasterized and a polygonal form. The aim is to provide a tool that mixes in a unified framework acquired discrete information (photos, MRI, ...) and synthetic continuous information (modeled objects). We propose the use of a multi-level hierarchical structure in which consecutive levels are linked. Each level of this structure corresponds to a particular representation of a same object. The lowest level is the raster representation and the highest level is the polygonal one. The way consecutive levels are computed, as well as the links between them are presented. Finally, we discuss the way the structure is updated using existing links in case of a modification applied on one level.
Rebollo,C., Remolar,I., Chover,M., Ripollés,O.
Abstract:
Outdoor scenes require vegetation to make them look realistic. Current hardware cannot afford real-time rendering of these scenes because of the large number of polygons. Multiresolution modelling has been successfully presented as a solution to the problem of efficient manipulation of highly detailed polygonal surfaces. This article describes a new continuous multiresolution hardware-oriented model that can represent tree foliage with different levels of detail. The multiresolution model presented in this paper, \emph{Level of Detail Foliage}, takes advantage of the programmable rendering pipelines nowadays available in most video cards. The geometry of the foliage is divided into a number of clusters, in some of which the detail can change while the rest of the clusters remain unaltered. This division of the foliage remarkably diminishes the number of vertices sent to the graphics system because only the information of the changed clusters are updated. The independent clusters condition a data structure that makes the time required for visualisation of the foliage more efficient. Here we present the data structure and the retrieval algorithms, which favour the extraction of an appropriate level of detail for rendering.
Kawaharada,H., Sugihara,K.
Abstract:
This paper proposes a new class of subdivision schemes. Previous subdivision processes are described by the movement and generation of vertices, and the faces are specified indirectly as polygons defined by those vertices. In the proposed scheme, on the other hand, the subdivision process is described by the generation of faces, and the vertices are specified indirectly as the intersections of these faces. In this sense, this paper gives a framework for a wide class of new subdivision methods. In short, the new subdivision is a dual framework of an ordinary subdivision based on the principle of duality in projective geometry. So, the new subdivision scheme inherits various properties of the ordinary subdivision schemes. In this paper, we define the dual subdivision and derive its basic properties.
Vanaken,C., Mertens,T., Bekaert,P.
Abstract:
Video-based rendering is a viable alternative to traditional realistic image synthesis techniques. It avoids the burden of time-consuming modeling and expensive global illumination simulation. In this paper we propose a video-based synthesis and animation system for fixed viewpoint scenes that feature rigid objects. We examplify this by using traffic video sequences. In a first stage, we extract vehicles and their trajectories from the example footage using an intuitive semi-automatic segmentation technique. Subsequently, the resulting vehicle ``sprites'' are dimensionally reduced using PCA in order to effectively extrapolate trajectories from incomplete sequences. Background, occlusions and shadows cast by vehicles are extracted as well. We show that convincing new footage can be synthesized readily from a single input video. Any number and variety of cars can be inserted, and their trajectories can be edited to simulate such traffic scenarios as lane changes and traffic jams.
Tobler,R., Maierhofer,S.
Abstract:
The photon map algorithm provides a number of advantages for fast global illumination algorithms. It calculates the direction independet illumination at a point in a scene, by estimating the local photon density. Standard estimation methods, which are based on computing the diameter of the sphere or ellipsoid containing the closest n photons, will have problems with typical architectural scenes, as walls and corners will lead to undesirable light and shadow leaks.
We introduce a new method for estimating photon density in photon maps, that is especially suited for calculating global illumination in architectural scenes and is based on two contributions: by using eight-sided, two-dimensional bounding box variants a better estimate for the area covered by the closest n photons can be computed. By providing additional information using a limited number of ray-casts, shadow and light leaks in the vicinity of walls can be significantly reduced. Combining these contributions results in a significant improvement in the speed and accuracy of global illumination algorithms based on photon maps.
Beets,K., Van Laerhoven,T.
Abstract:
While paint systems have been around for a long time, systems capable of capturing the complex behavior of paint media like watercolor, gouache, eastern ink, oil and acrylic paint have emerged only recently. However, concentrating on the simulation of paint and brush mechanics, these applications mostly provide a minimal set of instruments assisting users creating artwork. We report on the extension of our physically-based painting system for watery paint with a set of versatile tools supplying users with more control during the painting process. We introduce, among others, the use of masking pigment, a special-purpose brush using patterns to steer paint diffusion, and the adoption of an absorbent, textured piece of paper to remove some wet paint from the canvas. Results show that images created with genuine paint, using real-life counterparts of some of these tools, can be closely reproduced with our application. Additionally, our digital tools can produce effects that are difficult or impossible to achieve with real paint, while retaining the spontaneous nature of the resulting images.
Wüthrich,C.A., Augusto,J., Banisch,S., Wetzstein,G., Musialski,P.,Hofmann,T.
Abstract:
Children television productions have been using puppets for a long time, because children's attention is more drawn visually to colourful puppet characters on television than to regular adult television. Since the early days of computer animation, computer puppet simulation has been researched intensively. With the appearance of complex motion capture equipment, real time mapping of movement is now possible for virtual puppets (performance animation). However, the costs of capturing equipment are too high and the difference in the workflow make it difficult for small production teams to access and use such technology. This paper presents a system for the real time simulation of elastic latex puppets which are used in television productions. After an analysis of the production processes of real puppets and of the materials used for their production, the paper describes the components of the system simulating them. The main new feature of the system is the fact that it is capable to simulate three-dimensional elastic materials in real time, and to connect the elastic structure to a low cost data glove. This maps in a natural way the hand movements of a puppeteer to the computer model. The tradeoffs of the implementation on low cost hardware and its efficiency are also discussed.
Castro, S., Castro, L., Boscardín, L., De Giusti, A.
Abstract:
In this paper we propose a wavelet based model for large irregular volume data sets by exploiting a multiresolution model based on semi-regular tetrahedral meshes. In order to generate the multiresolution representation we use a wavelet based approach that allows compression and progressive transmission. Beginning with a semi-regular tetrahedral mesh Г∞(T) and applying the wavelet transform, we obtain a representation that consists of a coarse base mesh Г<span class="style2">0</span>(T) and a sequence of detail coefficients obtained from the sucessive decomposition of the mesh at different levels of resolution. The base mesh is the one at the lowest resolution and it does not have the connectivity subdivision property. The wavelet decomposition is obtained by defining a wavelet basis over tetrahedra generated by a regular subdivision method applied to an initial tetrahedron T. The obtained basis is a Haar-like one and forms an inconditional basis for Lp(T,A,m), 1<p<∞, being m the Lebesgue measure and A the sigma-algebra generated from the tetrahedron T by the chosen subdivision method.
Oore,S., Akiyama,Y.
Abstract:
We present a system that generates arm motion to new music tracks in MIDI format, based on sample motion-captured data of dancing to other pieces of music. However, rather than adapting existing motion, as most music-animation systems do, our system is novel in that it analyzes both the supplied motion and music data for certain characteristics (eg. melodic contour, loudness, etc), and learns relationships between the two. When new music is provided, the characteristics are analyzed as before, and used to predict characteristics of the motion. A generative process then creates motion curves according to these constraints. A demonstration is presented showing the results on both slow and fast tunes, including an automatically-generated trio of backup "dancers" (arm-motions only) for a jazz standard.
Carrard,T., Juliachs,M.
Abstract:
Recent advances in graphics processor architecture and capabilities have made the development of fast and efficient unstructured volume rendering methods possible. These techniques can be classified into two roughly delimited categories: cell projection based methods and GPU raycasting algorithms. However, both approaches are subject to limitations, respectively due to the main memory-to-GPU bandwidth for the former and due to the GPU per-fragment computation speed and memory size for the latter. These potential bottlenecks can be particularly limiting for large-size datasets, such as the ones produced by large-scale numerical simulation. In this work, we describe an enhancement to the cell-projection rendering method, allowing us to specify each tetrahedron with only 4 vertices and their associated data. By using a point sprite primitive, instead of a set of 4 triangles, we significantly reduce the amount of data transferred from the main memory through the graphics port for each frame rendered. We evaluate the impact of the different rendering stages of our method on the overall frame rate.
Roard,N., Jones,M.W.
Abstract:
This paper describes a flexible visualization architecture based on software agents, which enables the abstraction and reuse of rendering strategies. Using a reification of the rendering environment, the system is able to add new rendering strategies (such as distributed rendering or progressive rendering) to an existing pipeline, without any modification of the other components (controls components, display components, rendering algorithms...). The ability of changing strategies on the fly leads to a better adaptability to runtime constraints. The system uses an agent-based graphic pipeline, where each agent/component can be located on different computers; communications between agents use xml/rpc and data stream in order to easily integrate existing code in the system. Agents can add specific behavior to graphic pipelines, such as saving environments to reuse them, adapt informations and knowledge from another pipeline, and generally modify and improve the entire system. Various visualization and control clients exist, enabling collaboration between platforms such as PDAs, Windows, Linux, MacOS X, and Web (using Java applets).
Stavrakakis,J., Lau,Z.J., Lowe,N., Takatsuka,M.
Abstract:
With the abundance of high performance personal computers, rendering thousands to millions of polygons per second is an inexpensive task. In recent years, there have been advances in networking technologies that have enabled applications to become distributed over a network and many applications require this functionality. These applications can range from driving a large display, collaborating over an internet, or supporting pervasive environments. Solutions currently exist in providing graphics over a network. However, they are usually targeted to satisfy particular problem domains or are otherwise difficult to adopt as applications require major adjustment. OpenGL $^{\textregistered}$ is a graphics drawing library. Although many applications have made use of this API, few provide direct interaction within networked environments. In this paper we present Lumino, a framework that enables graphics from an existing OpenGL application to become available to a dynamic heterogeneous network. What differentiates Lumino from prior work is that it provides this functionality to existing unmodified applications at a very low level and is capable of supporting flow control, quality and scalability. Moreover, it is targeted at wide adoption and will be released under a Free Software license.
Tsong-Wuu Lin and Chung-Shen Hung
Abstract:
In this paper, we propose an image retrieval approach based on Quadrant Motif Scan (QMS). Motif scans from segmented blocks inside an image are the primary notion to extract image features. We exploit recursive quadrant segmentation in images and stratify hierarchical regions for matching comparison. Regions in the same stratum hold an identical credit, which is used for similarity metric. For the sake of matching flexibility, a dynamic adjustment scheme of credit setting is offered. In this sense, a user can arbitrarily adjust the credit parameters to pursue better retrieval results. Besides, a peak inspection technique is also added in the QMS matching metric to enhance performance. This means can helpfully refine retrieval performance with trivial computational cost. Experimental results reveal that effectiveness and efficiency of QMS are comparable to the Motif Cooccurrence Matrix (MCM) method while QMS is competent to deal with image scaling.
Ku,D.Ch., Qin,S.-F., Wright,D.K.
Abstract:
This paper presents a novel method for interpreting overtracing freehand sketch. The overtracing strokes are interpreted as sketch content and are used to generate 2D geometric primitives. The approach consists of four stages: stroke classification, strokes grouping and fitting, 2D tidy-up with endpoint clustering and parallelism correction, and in-context interpretation. Strokes are first classified into lines and curves by a linearity test. It is followed by an innovative strokes grouping process that handles lines and curves separately. The grouped strokes are fitted with 2D geometry and further tidied-up with endpoint clustering and parallelism correction. Finally, the in-context interpretation is applied to detect incorrect stroke interpretation based on geometry constraints and to suggest a most plausible correction based on the overall sketch context. The interpretation ensures sketched strokes to be interpreted into meaningful output. The interface overcomes the limitation where only a single line drawing can be sketched out as in most existing sketching programs, meanwhile is more intuitive to the user.
Birkholz,H.
Abstract:
In this paper, algorithms for the simplification and reconstruction of large triangle meshes are described. The simplification process creates a edge-collapse hierarchy in external memory, which is used for online reconstruction. The hierarchy indices are renamed after simplification, in order to allow fast reconstructions and the hierarchy is extended with information for view-dependent rendering.
The simplification makes no restrictions with the production of the hierarchy, but produces the same hierarchy as In-Core algorithms. The amount of memory, which is used for the simplification is adjustable.
Ryutarou Ohbuchi, Yushin Hata
Abstract:
In this paper, we propose and evaluate a systematic approach for improving performance of 3D model retrieval by combining multiple shape descriptors. We explored two approaches for generating multiple, mutually independent, shape descriptors; (1) application of a (single-resolution) shape descriptor on a set of multiresolution shape models generated from a query 3D shape model, and (2) application of multiple, heterogeneous shape descriptors on the query 3D shape model. The shape descriptors are integrated via the linear combination of the distance values they produce, using either fixed or adaptive weights. Our experiment showed that both multiresolution and heterogeneous sets of shape descriptors are effective in improving retrieval performance. For example, by using the multiresolution approach, the R-precision of the SPRH shape descriptor by Wahl, et al, improved by 8%, from 29% to 37%. A combination of three heterogeneous shape descriptors achieved the R-precision of about 42%; this figure is about 5% better than the R-precision of 38% achieved by the Light Field Descriptor by Chen, et al., which is arguably the best single shape descriptor reported to date.
Papaioannou,G., Gaitatzes,A., Christopoulos,D.
Abstract:
Occlusion culling is a genre of algorithms for rapidly eliminating portions of three-dimensional geometry hidden behind other, visible objects prior to passing them to the rendering pipeline. In this paper, an extension to the popular shadow frustum culling algorithm is presented, which takes into account the fact that many planar occluders can be grouped into compound convex solids, which in turn can provide fewer and larger culling frusta and therefore more efficient elimination of hidden geometry. The proposed method combines planar and solid occluders using a unified selection approach and is ideal for dynamic environments, as it does not depend on pre-calculated visibility data. The solid occluders culling algorithm has been applied to commercially deployed virtual reality systems and test cases and results are provided from actual virtual reality shows.
Szécsi,L.
Abstract:
We present an improved algorithm based on the Ray Engine approach, which builds a hierarchy of rays instead of objects, completely on the graphics card. Exploiting the coherence between rays when displaying refractive objects or computing caustics, realtime
frame rates are achieved without preprocessing. Thus, the method fills a gap in the realtime rendering repertoire.
Qin,S.F.
Abstract:
This paper presents a novel constructive approach to sequentially tidying up 2D online freehand sketches for further 3D interpretation in a conceptual design system. Upon receiving a sketch stroke, the system first identifies it as a 2D primitive and then automatically infers its 2D geometric constraints related to previous 2D geometry (if any). Based on recognized 2D constraints, the identified geometry will be modified accordingly to meet its constraints. The modification is realized in one or two sequent geometric constructions in consistence with its degrees of freedom. This method can produce 2D configurations without iterative procedures to solve constraint equations. It is simple and easy to use for a real-time application. Several examples are tested and discussed.
Stachera,J., Rokita,P.
Abstract:
Texture mapping is a technique for adding visual realism to the computer generated images. As the level of realism increases with the number and the resolution of textures, we are faced with the problem of limited texture memory space. Moreover, in order to alleviate the aliasing artefacts many graphics systems use the mip-mapping technique which needs to store additionally the texture pyramid. We propose an algorithm for texture compression which is characterized by low computational complexity, random access to compressed data and the hierarchical texture representation. The proposed hierarchical texture compression algorithm (HiTC) is based on a block-wise approach, where each block is subject to the modified fractal compression method and is partly represented by Laplacian pyramid. This allows us to incorporate the mip-map structure into the compressed texture and perfectly suits for real-time computer graphics applications.
Markovic,D., Gelautz,M.
Abstract:
In this paper, we present an algorithm to depict motion in comics-like form, with an artistic drawn-like representation of the scene, from a stereo image sequence. The input to the algorithm is a natural scene, along with a user-defined set of parameters that define the tone and stylistic properties of the image to be produced. The algorithm uses a dense disparity map, computed from the input stereo video, to preserve the perspective perception of the stylized image by drawing each stroke in a direction determined by the stereo derived disparity layers. To outline important features in the image, we utilize the contour edges provided by the Edge Combination algorithm. In the next step, we detect motion of the objects in the scene by tracking points that are close to the dominant edges of the Edge Combination image. The extracted dominant structure is further used to obtain a larger variety of styles for visualizing the motion trajectories. The output of the algorithm is a drawn-like form of the original scene with the motion highlighted in imitation of comics produced by hand.