Publication: Evaluación y aceleración del algoritmo Path tracing en arquitecturas heterogéneas
Loading...
Official URL
Full text at PDC
Publication Date
2021
Authors
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
En este trabajo se explica la implementación de Eleven Renderer, un motor de renderizado gráfico de mi autoría, programado en C++ y CUDA desde cero para este TFG. El código fuente se puede encontrar en el repositorio Github https://github.com/101001000/tfg-pathtracer/. Cuenta con todas las características necesarias para visualizar una escena 3D con relativa eficiencia y hace uso de la arquitectura paralela que ofrecen los aceleradores gráficos. Además se acompaña de una evaluación junto a la justificación de algunas decisiones tomadas durante el desarrollo. A lo largo del trabajo se exponen recursos en caso de querer ampliar información. Así pues, se espera que este trabajo sirva como breve guía y motivación para futuros programadores gráficos.
En este documento también se incluyen las mejoras visuales y optimizaciones aplicadas, junto a varias estadísticas de rendimiento comparando diferentes parámetros. Finalmente se ha puesto a disposición del lector la documentación necesaria para poder crear escenas 3D y visualizarlas de manera fotorrealista con Eleven Renderer, gracias a la posibilidad de importar de manera manual, escenas desde la gran mayoría de software de edición 3D.
In this work it's explained the implementation of Eleven Renderer, a rendering engine made by me from scratch, coded in C++ and CUDA library, done for this Bachelor Thesis. The source code can be found in Github's repo https://github.com/101001000/tfg-pathtracer/. It has all the features needed for 3D scene visualization with relative efficiency and it takes advantage of the parallel architecture which graphics accelerators offer. There's also an evaluation for the implementation and the justification of some decisions taken. Some resources are provided in case the reader wants in-depth insight about the topic. That being said, it's expected from this work to be used as a small guide and motivation for future developers and graphics programmers. This document also included some visual upgrades and optimizations, accompanied with benchmarks comparing different parameters. It's also offered to the reader, the documentation needed for creating 3D scenes and rendering it in a photorealistic way in Eleven Renderer, thanks to the possibility of manually importing scenes from another 3D editing software.
In this work it's explained the implementation of Eleven Renderer, a rendering engine made by me from scratch, coded in C++ and CUDA library, done for this Bachelor Thesis. The source code can be found in Github's repo https://github.com/101001000/tfg-pathtracer/. It has all the features needed for 3D scene visualization with relative efficiency and it takes advantage of the parallel architecture which graphics accelerators offer. There's also an evaluation for the implementation and the justification of some decisions taken. Some resources are provided in case the reader wants in-depth insight about the topic. That being said, it's expected from this work to be used as a small guide and motivation for future developers and graphics programmers. This document also included some visual upgrades and optimizations, accompanied with benchmarks comparing different parameters. It's also offered to the reader, the documentation needed for creating 3D scenes and rendering it in a photorealistic way in Eleven Renderer, thanks to the possibility of manually importing scenes from another 3D editing software.
Description
Trabajo de Fin de Grado en Ingeniería Informática, Facultad de Informática UCM, Departamento de Arquitectura de Computadores y Automática, Curso 2020-2021.