Title: Fast indirect lighting approximations using the representative candidate line space
Authors: Keul, Kevin
Koß, Tilman
Müller, Stefan
Citation: Journal of WSCG. 2018, vol. 26, no. 1, p. 11-20.
Issue Date: 2018
Publisher: Václav Skala - UNION Agency
Document type: článek
URI: wscg.zcu.cz/WSCG2018/!_2018_Journal_WSCG-No-1.pdf
ISSN: 1213-6972 (print)
1213-6980 (CD-ROM)
1213-6964 (on-line)
Keywords: vizualizace;počítačová grafika;trasování paprsku;datové struktury;algoritmy viditelnosti
Keywords in different language: visualization;computer graphics;ray tracing;data structures;visibility algorithms
Abstract in different language: We propose a novel approach for using directional Line Space information in calculation of indirect illumination. Typically, the Line Space is build on top of regular recursive grids and contains visibility information which is used to perform an efficient empty space skipping during traversal. In our method we extend the stored information by precomputed representative candidates, which are based on the Line Space shafts and serve as an approximation of the actual scene geometry. By using these candidates it is not necessary to compute any intersection tests and therefore the traversal is accelerated. However, the candidate approximation leads to visible artifacts. We therefore present a technique that significantly reduces these artifacts by extrapolation of the actual surface and demonstrate that the artifacts are nearly not perceivable in the application of indirect illumination. Moreover we adapt the Line Space to other data structures like bounding volume hierarchies (BVHs) which further increases the performance in ray tracing. Compared to the pure data structures we achieve significantly better performance with nearly no drawback in quality of indirect lighting.
Rights: © Václav Skala - UNION Agency
Appears in Collections:Volume 26, Number 1 (2018)

Files in This Item:
File Description SizeFormat 
Keul.pdfPlný text9,05 MBAdobe PDFView/Open

Please use this identifier to cite or link to this item: http://hdl.handle.net/11025/34584

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.