No description available.

Real-time rendering of massive 3D models has been recognized as a challenging task due to the limited computational power and memory available in a workstation. Most existing acceleration techniques, such as mesh simplification algorithms with hierarchical data structures, suffer from the nature of sequential executions. As data complexity increases due to the fundamental advances in modeling and simulation technologies, 3D models become complex and require gigabytes in storage. Consequently, visualizing such large datasets becomes a computationally intensive process where sequential solutions are unable to satisfy the demands of real-time rendering. Recently, the Graphics Processing Unit (GPU) has been praised as a massively parallel architecture not only for its significant improvements in performance but also because of its programmability for general-purpose computation. Today's GPUs allow researchers to solve problems by delivering fine-grained parallel implementations. In this dissertation, I concentrate on the design of parallel algorithms for real-time rendering of massive 3D polygonal models towards modern GPU architectures. As a result, the delivered rendering system supports high-performance visualization of 3D models composed of hundreds of millions of polygons on a single commodity workstation. / Ph. D.

Made available in DSpace on 2014-12-17T14:55:20Z (GMT). No. of bitstreams: 1 IcaroLLC_TESE.pdf: 6782954 bytes, checksum: cb0650b733e29c9e5e7032cef110a519 (MD5) Previous issue date: 2014-01-22 / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior / We revisit the problem of visibility, which is to determine a set of primitives potentially visible in a set of geometry data represented by a data structure, such as a mesh of polygons or triangles, we propose a solution for speeding up the three-dimensional visualization processing in applications. We introduce a lean structure , in the sense of data abstraction and reduction, which can be used for online and interactive applications. The visibility problem is especially important in 3D visualization of scenes represented by large volumes of data, when it is not worthwhile keeping all polygons of the scene in memory. This implies a greater time spent in the rendering, or is even impossible to keep them all in huge volumes of data. In these cases, given a position and a direction of view, the main objective is to determine and load a minimum ammount of primitives (polygons) in the scene, to accelerate the rendering step. For this purpose, our algorithm performs cutting primitives (culling) using a hybrid paradigm based on three known techniques. The scene is divided into a cell grid, for each cell we associate the primitives that belong to them, and finally determined the set of primitives potentially visible. The novelty is the use of triangulation Ja 1 to create the subdivision grid. We chose this structure because of its relevant characteristics of adaptivity and algebrism (ease of calculations). The results show a substantial improvement over traditional methods when applied separately. The method introduced in this work can be used in devices with low or no dedicated processing power CPU, and also can be used to view data via the Internet, such as virtual museums applications / N?s revisitamos o problema de visibilidade, que visa determinar um conjunto de primitivas potencialmente vis?veis em um conjunto de dados geom?tricos representados por uma estrutura de dados, por exemplo uma malha de pol?gonos ou de tri?ngulos, propondo uma solu??o para acelerar o processamento em aplica??es em visualiza??o tridimensional. Introduzimos uma estrutura enxuta, no sentido de abstra??o e redu??o de dados, que pode ser usada para aplica??es online e interativas. O problema de visibilidade ? especialmente importante na visualiza??o 3D de cenas representadas por grande volume de dados, em que n?o ? interessante manter todos os pol?gonos da cena em mem?ria. Isso implicaria em um maior tempo gasto na renderiza??o, ou sendo at? mesmo imposs?vel mant?-los todos em volumes imensos de dados. Nestes casos, dada uma posi??o e uma dire??o de visualiza??o, o objetivo principal ? determinar e carregar o m?nimo poss?vel de primitivas (pol?gonos) da cena, visando acelerar a etapa de renderiza??o. Para este prop?sito, nosso algoritmo executa o corte de primitivas (culling) usando um paradigma h?brido baseado em tr?s modelos conhecidos. A cena ? subdividida em c?lulas de uma grade, sendo associada a cada uma dessas c?lulas as primitivas pertencentes a elas, e finalmente determinado o conjunto de primitivas potencialmente vis?veis. A novidade ? o uso da triangula??o J a 1 para criar a subdivis?o em grade. Escolhemos esta estrutura devido ?s suas caracter?sticas relevantes de adaptatividade e algebrismo (facilidade de c?lculos). Os resultados mostram uma melhoria substancial sobre os m?todos tradicionais quando aplicados separadamente. O m?todo introduzido neste trabalho pode ser usado em dispositivos sem processador dedicado ou com baixo poder de processamento, e ainda, pode ser utilizado para visualizar dados atrav?s da Internet, tal como em aplica??es de museus virtuais