A Methodical Approach to the Evaluation of Light Transport Computations
Název práce v češtině: | Metodický přístup k evaluaci výpočtů transportu světla |
---|---|
Název v anglickém jazyce: | A Methodical Approach to the Evaluation of Light Transport Computations |
Klíčová slova: | počítačová grafika, renderování, simulace transportu světla, globální osvětlení, testovací scény, metriky kvality obrazu |
Klíčová slova anglicky: | computer graphics, rendering, light transport simulation, global illumination, test scenes, image quality metrics |
Akademický rok vypsání: | 2019/2020 |
Typ práce: | diplomová práce |
Jazyk práce: | angličtina |
Ústav: | Katedra softwaru a výuky informatiky (32-KSVI) |
Vedoucí / školitel: | doc. Alexander Wilkie, Dr. |
Řešitel: | skrytý - zadáno a potvrzeno stud. odd. |
Datum přihlášení: | 27.03.2020 |
Datum zadání: | 27.03.2020 |
Datum potvrzení stud. oddělením: | 29.04.2020 |
Datum a čas obhajoby: | 16.09.2020 09:00 |
Datum odevzdání elektronické podoby: | 30.07.2020 |
Datum odevzdání tištěné podoby: | 30.07.2020 |
Datum proběhlé obhajoby: | 16.09.2020 |
Oponenti: | Dipl.-Ing. Ivo Kondapaneni |
Zásady pro vypracování |
Photorealistic rendering has a wide variety of applications, and so there are many different algorithms and their variations tailored for specific use cases. Sometimes they are optimized for specific purposes at the expense of others, or they generate biased results or do fast approximations. But practically all of them are based on physically correct simulations of light transport - usually some form of path tracing. Therefore all of them generate photorealistic images within some specified margin of error (for a given algorithm), and their results on the same scene will likely be somewhat different for each algorithm.
If we want to select the best algorithm for a specific task, we need to compare all the possibilities. That is quite difficult because there is no standardized testing available. Specifically, properties of the light transport algorithms are usually presented on different sets of 3D scenes of varying kinds and complexities. Therefore the only way to get an unbiased comparison of multiple algorithms is to implement them and test them on the same set of scenes. That is not often feasible because these algorithms are fairly complex and their implementation difficult and time-consuming. For these reasons, the goal of the thesis is to develop an extensible test suite of light transport algorithms which is made available for as many popular rendering frameworks as possible. Specifically, we need to develop a curated set of canonical test scenes based on both purely mathematical understanding of light transport theory, and typical technical issues which arise during the implementation of modern light transport algorithms. Together, these scenes will be a good representation of feature space with regards to light transport, and both their design and properties must be exactly documented. Additionally, software for automated benchmarking and presentation of its results in the form of web pages will be provided. With this thesis, we want to provide a foundation for a standardized testing framework of light transport algorithms, which is currently missing in the field of photorealistic rendering. |
Seznam odborné literatury |
Ulbricht, Christiane & Wilkie, Alexander & Purgathofer, Werner. (2006). Verification of Physically Based Rendering Algorithms. Comput. Graph. Forum. 25. 237-255. 10.1111/j.1467-8659.2006.00938.x.
Pharr, Matt, Wenzel Jakob and Greg Humphreys. Physically based rendering: from theory to implementation. Third edition. Cambridge, MA: Morgan Kaufmann Publishers/Elsevier, [2017]. ISBN 978-0-12-800645-0. Jakob, Wenzel. “Mitsuba renderer”. http://www.mitsubarenderer.org (2010). URL: http://www.mitsuba-renderer.org Merlin Nimier-David, Delio Vicini, Tizian Zeltner, and Wenzel Jakob. 2019. Mitsuba 2: A Retargetable Forward and Inverse Renderer. In Transactions on Graphics (Proceedings of SIGGRAPH Asia) 38(6). |