Efficient Bidirectional Path Tracing by Randomized Quasi-Monte Carlo Integration

  • As opposed to Monte Carlo integration the quasi-Monte Carlo method does not allow for an (consistent) error estimate from the samples used for the integral approximation. In addition the deterministic error bound of quasi-Monte Carlo integration is not accessible in the setting of computer graphics, since usually the integrands are of unbounded variation. The structure of the high dimensional functionals to be computed for photorealistic image synthesis implies the application of the randomized quasi-Monte Carlo method. Thus we can exploit low discrepancy sampling and at the same time we can estimate the variance. The resulting technique is much more efficient than previous bidirectional path tracing algorithms.

Export metadata

Additional Services

Search Google Scholar
Metadaten
Author:Thomas Kollig, Alexander Keller
URN:urn:nbn:de:hbz:386-kluedo-49770
Series (Serial Number):Interner Bericht des Fachbereich Informatik (310)
Document Type:Report
Language of publication:English
Date of Publication (online):2017/10/27
Year of first Publication:2001
Publishing Institution:Technische Universität Kaiserslautern
Date of the Publication (Server):2017/10/27
Page Number:15
Faculties / Organisational entities:Kaiserslautern - Fachbereich Informatik
DDC-Cassification:0 Allgemeines, Informatik, Informationswissenschaft / 004 Informatik
Licence (German):Creative Commons 4.0 - Namensnennung, nicht kommerziell, keine Bearbeitung (CC BY-NC-ND 4.0)