<p>Real-time global illumination has been a very important topic and is widely used in game industry. Previous offline rendering requires a large amount of time to converge and reduce the noise generated in Monte Carlo method. Thus, it cannot be easily adapted in real-time rendering. Using voxels in the field of global illumination has become a popular approach. While a naïve voxel grid occupies huge memory in video card, a data structure called <i>sparse voxel octree</i> is often implemented in order to reduce memory cost of voxels and achieve efficient ray casting performance in an interactive frame rate. </p> <p>However, rendering of voxels can cause block effects due to the nature of voxel. One solution is to increase the resolution of voxel so that one voxel is smaller than a pixel on screen. But this is usually not feasible because higher resolution results in higher memory consumption. Thus, most of the global illumination methods of SVO (sparse voxel octree) only use it in visibility test and radiance storage, rather than render it directly. Previous research has tried to incorporate SVO in ray tracing, radiosity methods and voxel cone tracing, and all achieved real-time frame rates in complex scenes. However, most of them only focus on static scenes and does not consider dynamic updates of SVO and the influence of it on performance.</p> <p>In this thesis, we will discuss the tradeoff of multiple classic real-time global illumination methods and their implementations using SVO. We will also propose an efficient approach to dynamic update SVO in animated scenes. The deliverables will be implemented in CUDA 11.0 and OpenGL.</p>

Background. Global illumination is not trivial to compute in real-time computer graphics. One approximate solution is to distribute virtual light sources from a primary light, to then apply direct light calculations to said virtual lights. This can effectively estimate two-bounce illumination. To mitigate artifacts, virtual lights make use of a spherical shape and utilize spherical harmonics to allow for efficient light integration. These indirect light sources are referred to as "harmonics virtual lights" (HVLs). Objectives. The objectives of this thesis are to analyze the data structures, calculations and performance of an HVL implementation in different 3D scenes. Methods. HVLs are implemented using the Vulkan API. Experiments are then performed to evaluate and optimize execution times. Furthermore, different measures are taken to ensure correctness and minimize errors wherever possible. Results. The GPU pass responsible for gathering indirect light contributions from HVLs turned into a heavy bottleneck. A number of different optimization techniques were applied to said pass and analyzed. Seven techniques were found to have a positive effect on performance, each with varying degrees of impact on timings. No optimization compromised on input parameters, visual results or mathematical correctness. Additionally, three techniques were instead worsening performance of the implementation, despite having initial motivations for possible improvements. Conclusions. All optimization techniques with positive effects working in conjunction led to a total speedup of 46.9x in a specific use case of our implementation. There is room for further potential improvements, and a number of different techniques for future work are explained. The final source code for the implementation can be viewed in a public GitHub repository.

The interaction of light in our world is immensely complex, but with mod- ern computers and advanced rendering algorithms, we are beginning to reach the point where photo-realistic renders are truly difficult to separate from real photographs. Achieving realistic or believable global illumination in scenes with participating media is exponentially more expensive compared to our traditional polygonal methods. Light interacts with the particles of a volume, creating com- plex radiance patterns. In this thesis, we introduce an extension to the commonly used point-based color bleeding (PCB) technique, implementing volume scatter contributions. With the addition of this PCB algorithm extension, we are able to render fast, be- lievable in- and out-scattering while building on existing data structures and paradigms. The proposed method achieves results comparable to that of existing Monte Carlo integration methods, obtaining render speeds between 10 and 36 times faster while keeping memory overhead under 5%.

Modeling believable lighting is a crucial component of computer graphics applications, including games and modeling programs. Physically accurate lighting is complex and is not currently feasible to compute in real-time situations. Therefore, much research is focused on investigating efficient ways to approximate light behavior within these real-time constraints. In this thesis, we implement a general purpose algorithm for real-time applications to approximate indirect lighting. Based on voxel cone tracing, we use a filtered representation of a scene to efficiently sample ambient light at each point in the scene. We present an approach to scene voxelization using hardware tessellation and compare it with an approach utilizing hardware rasterization. We also investigate possible methods of warped voxelization. Our contributions include a complete and open-source implementation of voxel cone tracing along with both voxelization algorithms. We find similar performance and quality with both voxelization algorithms.

Creation of visually-pleasing images has always been one of the main goals of computer graphics. Two important components are necessary to achieve this goal --- artists who design visual aspects of an image (such as materials or lighting) and sophisticated algorithms that render the image. Traditionally, rendering has been of greater interest to researchers, while the design part has always been deemed as secondary. This has led to many inefficiencies, as artists, in order to create a stunning image, are often forced to resort to the traditional, creativity-baring, pipelines consisting of repeated rendering and parameter tweaking. Our work shifts the attention away from the rendering problem and focuses on the design. We propose to combine non-physical editing with real-time feedback and provide artists with efficient ways of designing complex visual aspects such as global illumination or all-frequency shadows. We conform to existing pipelines by inserting our editing components into existing stages, hereby making editing of visual aspects an inherent part of the design process. Many of the examples showed in this work have been, until now, extremely hard to achieve. The non-physical aspect of our work enables artists to express themselves in more creative ways, not limited by the physical parameters of current renderers. Real-time feedback allows artists to immediately see the effects of applied modifications and compatibility with existing workflows enables easy integration of our algorithms into production pipelines.

La génération d'images de synthèse pour la production cinématographique n'a cessé d'évoluer durant ces dernières décennies. Pour le non-expert, il semble que les effets spéciaux aient atteint un niveau de réalisme ne pouvant être dépassé. Cependant, les logiciels mis à la disposition des artistes ont encore du progrès à accomplir. En effet, encore trop de temps est passé à attendre le résultat de longs calculs, notamment lors de la prévisualisation d'effets spéciaux. La lenteur ou la mauvaise qualité des logiciels de prévisualisation pose un réel problème aux artistes. Cependant, l'évolution des cartes graphiques ces dernières années laisse espérer une potentielle amélioration des performances de ces outils, notamment par la mise en place d'algorithmes hybrides rasterisation/ lancer de rayons, tirant profit de la puissance de calcul de ces processeurs, et ce, grâce à leur architecture massivement parallèle. Cette thèse explore les différentes briques logicielles nécessaires à la mise en place d'un pipeline de rendu complexe sur carte graphique, permettant une meilleure prévisualisation des effets spéciaux. Différentes contributions ont été apportées à l'entreprise durant cette thèse. Tout d'abord, un pipeline de rendu hybride a été développé (cf. Chapitre 2). Par la suite, différentes méthodes d'implémentation de l'algorithme de Path Tracing ont été testées (cf. Chapitre 3), de façon à accroître les performances du pipeline de rendu sur GPU. Une structure d'accélération spatiale a été implémentée (cf. Chapitre 4), et une amélioration de l'algorithme de traversée de cette structure sur GPU a été proposée (cf. Section 4.3.2). Ensuite, une nouvelle méthode de décorrélation d'échantillons, dans le cadre de la génération de nombres aléatoires a été proposée (cf. Section 5.4) et a donné lieu à une publication [Dufay et al., 2016]. Pour finir, nous avons tenté de combiner l'algorithme de Path Tracing et les solutions Many Lights, toujours dans le but d'améliorer la prévisualisation de l'éclairage global. Cette thèse a aussi donné lieu à la soumission de trois mémoires d'invention et a permis le développement de deux outils logiciels présentés en Annexe A. / Image synthesis for movie production never stopped evolving over the last decades. It seems it has reached a level of realism that cannot be outperformed. However, the software tools available for visual effects (VFX) artists still need to progress. Indeed, too much time is still wasted waiting for results of long computations, especially when previewing VFX. The delays or poor quality of previsualization software poses a real problem for artists. However, the evolution of graphics processing units (GPUs) in recent years suggests a potential improvement of these tools. In particular, by implementing hybrid rasterization/ray tracing algorithms, taking advantage of the computing power of these processors and their massively parallel architecture. This thesis explores the different software bricks needed to set up a complex rendering pipeline on the GPU, that enables a better previsualization of VFX. Several contributions have been brought during this thesis. First, a hybrid rendering pipeline was developed (cf. Chapter 2). Subsequently, various implementation schemes of the Path Tracing algorithm have been tested (cf. Chapter 3), in order to increase the performance of the rendering pipeline on the GPU. A spatial acceleration structure has been implemented (cf. Chapter 4), and an improvement of the traversal algorithm of this structure on GPU has been proposed (cf. Section 4.3.2). Then, a new sample decorrelation method, in the context of random number generation was proposed (cf. Section 5.4) and resulted in a publication [Dufay et al., 2016]. Finally, we combined the Path Tracing algorithm with the Many Lights solution, always with the aim of improving the preview of global illumination. This thesis also led to the submission of three patents and allowed the development of two software tools presented in Appendix A.

Le rendu réaliste est devenu essentiel dans l'industrie (cinéma, jeux vidéo, prototypage et design...). Cela nécessite de simuler l'interaction entre la lumière et les objets d'une scène 3D, un calcul connu sous le nom d'Illumination Globale et habituellement très coûteux en temps de calcul. Nous présentons une technique d'illumination globale combinant deux méthodes usuelles : les cartes de photons et le cache d'éclairement. Les cartes de photons ne dépendent pas de la vue, mais nécessitent une passe coûteuse appelée regroupement final. Le cache d'éclairement est plus rapide mais dépend de la vue : pour couvrir la scène entière, l'utilisateur doit placer manuellement plusieurs caméras dans la scène. Notre méthode exploite les avantages de chaque méthode, sans intervention de l'utilisateur. Elle génère un cache d'éclairement de qualité indépendant de la vue à partir d'une carte de photons, affichable interactivement. Nous étudions également une nouvelle approche pour réduire la variance inhérente aux méthodes de Monte Carlo. En règle générale, les emplacements des échantillons sont ignorés : deux échantillons proches se voient attribuer la même importance, bien qu'ayant probablement des valeurs similaires. L'approche bayésienne que nous proposons dans cette thèse utilise la valeur et la position des échantillons et se base sur un modèle probabiliste de l'intégrant pour inférer une valeur de l'intégrale. L'estimée bayésienne ne dépend que des échantillons, et non pas de la manière dont ils ont été choisis. Nous montrons que cette approche peut être appliquée au calcul du regroupement final et nous présentons des résultats démontrant l'intérêt du Monte Carlo Bayésien.

The objective of this thesis is the development of algorithms for the simulation of the light transport in general environments to render high quality still images. To this end, first we have analyzed the existing methods able to render participating media, more concretely those that account for multiple scattering within the media. Next, we have devised a couple of two pass methods for the computation of those images. For the first step we have proposed algorithms to cope with the scenes we want to deal with. The second step uses the coarse solution of the first step to obtain the final rendered image.The structure of the dissertation is briefly presented below.In the first chapter the motivation of the thesis and its objectives are discussed. It also summarizes the contributions of the thesis and its organization.In the second chapter the principles of global illumination for general environments are reviewed, with the most important equations---the rendering equation and the transport equation---whose solution constitutes the global illumination problem. In order to solve the global illumination problem, a certain number of multi-pass methods exist. Their objective is to be able to skip restrictions on the number of types of light paths that could be dealt with a single technique, or increase efficiency and/or accuracy. We have opted to follow this philosophy, and a pair of two pass methods have been developed for general environments.The third chapter includes the study of the methods that perform the single scattering approximation, and also the study of the ones that take into account multiple scattering.The fourth chapter is devoted to our first pass method, which computes a rough estimate of the global illumination. Knowing the benefits of hierarchical approaches, two concrete algorithms based on hierarchies have been extended to be more generic: Hierarchical Radiosity with Clustering and Hierarchical Monte Carlo Radiosity.Our second pass is considered in the next chapter. Using the coarse solution obtained by the first pass, our second pass computes a high quality solution from a given viewpoint. Radiances and source radiances are estimated using Monte Carlo processes in the context of path tracing acceleration and also for final gather. Probability density functions (PDFs) are created at ray intersection points. For such a task, we initially used constant basis functions for the directional domain. After realizing of their limitations we proposed the Link Probabilities (LPs), which are objects with adaptive PDFs in the links-space.In order to take advantage of the effort invested for the construction of the LPs, we have devised two closely related progressive sampling strategies. In the second pass, instead of sampling each pixel individually, only a subset of samples is progressively estimated across the image plane. Our algorithms are inspired by the work of Michael D. McCool on anisotropic diffusion using conductance maps.The final chapter presents the conclusions of the thesis. Also possible lines of further research are suggested. / El objetivo de esta tesis es el desarrollo de algoritmos para la simulación del transporte de la luz en los entornos genéricos para generar imágenes de la alta calidad. Con este fin, primero hemos analizado los métodos existentes capaces de visualizar medios participativos, más concretamente los que tienen en cuenta la dispersión múltiple en los medios. Después, hemos ideado un par de métodos de dos pasos para el cómputo de esas imágenes. Para el primer paso hemos propuesto algoritmos que hacen frente a las escenas que deseamos tratar. El segundo paso utiliza la solución aproximada del primer paso para obtener la imagen final. La estructura de la disertación se presenta brevemente en lo que sigue.En el primer capítulo se discuten la motivación de la tesis y sus objetivos. También se resumen las contribuciones de la tesis y su organización. En el segundo capítulo se repasan los principios de la iluminación global para los ambientes genéricos, con las ecuaciones-más importantes (la ecuación de rendering y la ecuación de transporte) cuya solución constituye el problema global de iluminación. Para solucionar el problema global de iluminación, cierto número de métodos de múltiples pasos existen. Su objetivo es poder eliminar restricciones en el número de tipos de caminos de luz que se podrían tratar con una sola técnica, o aumentar su eficacia y/o exactitud. Hemos optado seguir esta filosofía, desarrollando un par de métodos de dos pasos para entornos genéricos.El tercer capítulo incluye el estudio de los métodos que utilizan la aproximación de dispersión simple, y también el estudio de los que consideran la dispersión múltiple.El cuarto capítulo está dedicado a nuestro método de primer paso, que computa un cálculo aproximado de la iluminación global. Conociendo las ventajas de los métodos jerárquicos, dos algoritmos concretos basados en jerarquías se han ampliado para ser más genéricos: radiosidad jerárquica con clustering y radiosidad jerárquica usando Monte Carlo. Nuestro segundo paso se considera en el capítulo siguiente. Usando la solución aproximada obtenida por el primer paso, el segundo paso computa una solución de la alta calidad para un punto de vista dado. Se estiman las radiancias usando procesos de Monte Carlo en el contexto de la aceleración de trazadores de rayos y también para final gather. Las funciones de densidad de probabilidad (PDFs) se crean en los puntos de interacción de los rayos. Para tal tarea, utilizamos inicialmente funciones constantes como base para el dominio direccional. Después de comprender sus limitaciones, propusimos establecer probabilidades directamente sobre los enlaces (link probabilities, o LPs), usando objetos con PDFs adaptativos en el espacio de los enlaces.Para aprovechar el esfuerzo invertido en la construcción de los LPs, hemos ideado dos estrategias de muestreo progresivas. En el segundo paso, en vez de muestrear cada pixel individualmente, solamente se estima progresivamente un subconjunto de muestras a través del plano de imagen. Nuestros algoritmos han sido inspirados en el trabajo de Michael D. McCool en la difusión anisotrópica usando mapas de conductancia.El capítulo final presenta las conclusiones de la tesis, y también sugiere las líneas posibles de investigación futura.

[pt] Um dos principais objetivos da computação gráfica é a geração de imagens fotorrealistas, ou seja, imagens indistinguíveis das de uma capturada por uma câmera real ou, mais ambiciosamente, imagens que provocam a mesma sensação no sistema visual de um observador quando olhando diretamente para uma cena. Aplicações incluem o projeto de iluminação, a arquitetura, a realidade virtual e a indústria do cinema. Esse trabalho apresenta a arquitetura de um sistema capaz de resolver, por técnicas de Monte Carlo, a equação do transporte da luz, essencialmente uma aproximação das equações de Maxwell para ótica geométrica. Além de ser um renderer funcional, o sistema é implementado na forma de um framework em cima do qual pode-se facilmente experimentar idéas na área de síntese de imagens fotorrealistas, como, por exemplo, diferentes materiais, geometrias, estruturas de aceleração e estratégias de integração. / [en] One of the main goals in computer graphics is to create photorealistic images, that is, images indistinguishable from the ones captured by a real camera, or, more ambitiously, images that cause the same sensation on the visual system of an observer looking directly at a scene. Applications include illumination design, architecture, virtual reality and the movie industry. This work presents the architecture of a renderer that solves by Monte Carlo techniques the light transport equation, essentially a geometric optics approximation of Maxwell´s equations. The system is not only a functional renderer but also a framework where we can easily experiment new ideas in photorealistic image synthesis, such as new materials, geometries, acceleration structures and integration techniques.

BARRETO, Isaac Moreira. Cálculo do Fator-de-Forma exato entre Áreas Diferencial e Finita Usando CSG. 2008. 55 f. : Dissertação (mestrado) - Universidade Federal do Ceará, Centro de Ciências, Departamento de Computação, Fortaleza-CE, 2008. / Submitted by guaracy araujo (guaraa3355@gmail.com) on 2016-07-01T17:52:27Z No. of bitstreams: 1 2008_dis_imbarreto.pdf: 752627 bytes, checksum: 2c1a97d41785e527e97633cc6c7e9756 (MD5) / Approved for entry into archive by guaracy araujo (guaraa3355@gmail.com) on 2016-07-01T17:56:22Z (GMT) No. of bitstreams: 1 2008_dis_imbarreto.pdf: 752627 bytes, checksum: 2c1a97d41785e527e97633cc6c7e9756 (MD5) / Made available in DSpace on 2016-07-01T17:56:22Z (GMT). No. of bitstreams: 1 2008_dis_imbarreto.pdf: 752627 bytes, checksum: 2c1a97d41785e527e97633cc6c7e9756 (MD5) Previous issue date: 2008 / The Ray-Tracing and Radiosity methods are the main representatives of the method that solve the global illumination problem. In both mthods it is necessary to know the energy tranfer ratio between two areas. This ratio, called form factor, is one of the key concepts in Radiosity methods and is being more frequently used in Ray-Tracing methods with finite area light sources. There are many methods for the computation of the form factor, most of them are approximative due to a matter of performance, but, in some specific cases, the extra computational effort needed to compute the exact value of the form factor can improve the overall performance of the illumination method. In general, in these cases, the computational effort needed to obtain an acceptable approximation of the form factor outweighs the effort necessary to compute the exact value. Furthermore there are situation, for example, shadow boundary shading, in which a high precision is far more important than a performance gain. In this work we present a method to compute the exact form factor between a finite area and a differential area which uses CSG techniques to identify the ooccluded areas of the source. / Os métodos de Ray-Tracing e Radiosidade são os principais representantes dos métodos existentes para resolver o problema de iluminação global. Em ambos os métodos se faz necessário saber a taxa de transferência de energia luminosa entre duas áreas. Essa taxa de transferência, chamada de fator-de-forma, é um dos pontos principais no método de Radiosidade e vem sendo usado cada vez com mais frequência em métodos de Ray-Tracing com fontes luminosas de área finita. Existem vários métodos para o cálculo do fator-de-forma, a maioria deles são aproximativos por uma questão de desempenho. Porém, em casos específicos, o trabalho extra para calcular o valor exato do fator-de-forma pode melhorar o desempenho global do método. Em geral, nesses casos, o esforço necessário para se obter uma aproximação aceitável do valor do fator-de-forma supera o esforço necessário para calcular o valor exato em si. Além disso, existem situações, tais como a renderização nas áreas de fronteiras de sombras, em que uma alta precisão é mais importante do que um ganho no desempenho. Nessas situações, é desejável que o método tenha ao seu dispor uma maneira de calcular o valor exato do fator-de-forma. Neste trabalho é apresentado um método para calcular o fator-de-forma exato entre uma área finita e uma área diferencial que utiliza de técnicas CSG para identificar as áreas ocluídas do polígono emissor.