Catálogo de publicaciones - libros

Shadow maps provide a fast and convenient method of generating real-time shadows in computer games. Unfortunately, shadow maps suffer from aliasing problems. Percentage Closer Filtering (PCF) is one of the solutions, but in cases of severe aliasing it can only mask the aliasing by unnaturally blurring. In this paper we present two improved antialiased shadow algorithms based on PCF, named as Modified PCF I and Modified PCF II. The primary idea of our algorithms is to make sawtooth shadow edges straighter just by data modification before bilinear interpolation in PCF. We respectively implement the algorithms with graphic hardware in Pixel Shader 2.0 and 3.0, several examples are also provided to show that most of aliasing artifacts have been removed compared with PCF. The time complexity and hardware requirement of Modified PCF I are almost as same as that of PCF. The resource requirement of Modified PCF II is more rigorous than that of Modified PCF I, but it can still accommodate to real-time game rendering with obvious antialiasing effect.

In this paper, a new method is described for rendering convincing soft shadows from area light source at interactive frame rates. The method is based on the shadow map algorithm and it fully exploits the parallelism of the programmable graphics pipeline. In the method we sample the area light source using parallel linear segments and compute the light contribution of each segment using a completely new method which gives reasonable simulation for soft shadow generated by tiny scene object. We demonstrate that it can generate detailed soft shadow, while achieving the high frame rates. And our approach offers a number of benefits. E.g., it’s easy to be adopted by the projects using shadowing map and there is no extra overhead while the complexity of the scene geometry increases.

The navigation and rendering of very large-scale terrain are facing a difficult problem that the geometry data and texture data cannot be used directly due to the storage space, computation capacity, and I/O bandwidth. To provide more realistic detail of terrain scene, procedural detail is a good solution. Firstly, this paper introduces a method of procedural geometry based on the terrain tile quad-tree and the Patch-LOD algorithm. Then, the texture generation operator is described and the method of dynamic pre-computation of patch-texture is presented. Finally, the experimental system based on these above ideas and methods has been implemented. The experimental results show that these methods are effective and are appropriate to the development of 3D games and battlefield applications.

A uniform active model adaptation is proposed for some specific applications related with realistic face reconstruction and sketch generation, which is separated into tow phases: global adaptation and local adaptation. During the global adaptation, a coarse matching is used for adjusting the facial contour and five sense organs according to the correspondence of feature points both in 2D facial image and 3D model. An interpolation is processed on the global adjusted 3D model for a fine matching, which forms the local adaptation. A personalized facial sketch generation method is present based on the similar idea of 3D facial model adaptation. By comparison between the feature points of specific person’s face and those of a predefined generic face, metamorphosis can be drawn for generating a personalized cartoon facial sketch. Two systems are built, and face reconstruction and sketch results show that proposed method is efficient and desirable.

Taking Dunhuang MoGao Frescoes as research background, a new method to generate the sketch for Dunhuang frescoes is proposed. Learning from the traditional manual imitation procedure, we first get the contour line of a Dunhuang fresco which is composed of many connected curves, and then render the strokes by learning styles from examples. For several example strokes, the style of example strokes are captured and expressed as style function. Given a target curve, a new stroke can be generated with the style of one example and moreover, the style of the target stroke can be further customized into different styles similar to the examples by interpolation of styles. This research has the potential to provide a computer aided tool for art historians to do imitation work, and improve the efficiency of imitation.

We introduce a new surface deformation and modeling method in this paper. Referring to the swept volume generation, the surface is pulled or pushed along a trajectory curve. The key point is the sweeping function. Surface points are moved to where they should be during sweeping operations according to the global parameter, which is determined by topological distance. An index factor controls how much the surface deforms around the handle point. The proposed method is easy to extend to fit different applications such as various constraints, local deformation and animations.

With the development of motion capture techniques; more and more 3D motion libraries become available. The growing amount of motion capture data requires more efficient and effective methods for indexing, searching and retrieving. In many cases, the user will only have a sketchy idea of which kind of motion to look for in the motion database. In consequence, the description about the query movement is a bottleneck for motion retrieval system. This paper presents a framework that can describe and handle the query scenes effectively. Our content-based retrieval system supports two kinds of query modes: textual query mode and query-by-example mode. By using various kinds of qualitative features and adaptive segments of motion capture data stream, our indexing and retrieval methods are carried out at the segment level rather than at the frame level, making them quite efficient. Some experimental examples are given to demonstrate the effectiveness and efficiency of proposed algorithms.

In this paper, we present a fast individual face modeling and animation system. The input of the system is a single frontal face image, which is captured by an ordinary video camera. And the output is 3D individual facial animation. Firstly, an Adaboost method is used to detect the face in the image. Then the key points of the face are located with the inverse compositional algorithm based active appearance models (AAM). The 3D geometry mesh of the individual face is then generated with morphable models. Finally, the direction and magnitude of existed motion vectors are adjusted according to different 3D face shapes using expression cloning technique. With this system, we can achieve realistic 3D facial animation with minimal and often no manual tuning. The experiments show that our system is fast enough for common applications.

In this paper, we describe a haptic deformation algorithm used in a virtual storytelling system known as the Digital Buddhist Image Creation System. This system supports not only haptic interaction, but also visual and auditory feedback and the combination of these three modalities provides a user with an immersive experience of a virtual cultural environment. We believe the proposed bi-directional haptic interaction will encourage users to spontaneously participate and engage with the system. To demonstrate and explore this concept, a virtual environment that tells the legend of Korea’s ’Unju Temple’ has been constructed. The proposed haptic interaction is expected to play an important role in supporting users as they directly experience and learn about this significant piece of cultural heritage in a digital, virtual environment.

Emotion plays an essential role in adaptation and social communication in organism. Similarly, appropriately timed and clearly expressed emotion is a central required for believable interactive virtual humans. Presently, incorporating emotion into virtual humans is gain increasing attention in academia and industry. This strong interest is driven by a wide spectrum of promising applications in many areas such as virtual reality, e-learning, entertainment, etc.This paper discusses a system that allows the design of emotion for interactive virtual human. We introduce an emotion model of artificial psychology, in which the transition of emotion can be viewed as a Morkov process and the relation of between emotion and mood can be describes by Hidden Morkov Model (HMM). In order to demonstrate the approach, we integrate the emotion model into a system composed of voice recognition and a realistic facial model. Thus, the model can be used for generating a variety of emotional expression of autonomous, interactive virtual humans.