Mathematical Models and Monte Carlo Algorithms for Physically Based
Eric Lafortune, February 1996.
The dissertation gives an overview of Monte Carlo techniques and optimisations
for photo-realistic image synthesis. Various unbiased variance reduction
techniques are discussed and compared, both in theory and in practice. The
recent bidirectional path tracing algorithm is presented and compared to
classical path tracing.
Have a look at the following pages to see if there's anything in it that could
The entire text is also available on-line. At
Downloading the larger files seems to be problematic at the moment. If you're
on a distant site, you'll most certainly want to use one of the two faster
mirror sites. You can download the files from these sites by following the
- The cover pages
(2 pages, 28 K of gzip'ed Postscript, expands to 61 K)
- The dissertation with grey-scale images
(142+2 pages, 4.2 MB of gzip'ed Postscript, expands to 10.5 MB)
- A more compact version of the dissertation
with b/w images
(1.1 MB of gzip'ed Postscript, expands to 3.8 MB)
- The summary in Dutch
(21+2 pages, 126 K of gzip'ed Postscript, expands to 417 K)
- At ftp.funet.fi:
Thanks to Juhana Kouhia for putting them there.
- At www.graphics.cornell.edu:
Please let me know if you have any problems with viewing or printing any of
these files. They have been created using LaTeX, bibtex, dvips, gnuplot, make,
sed, Framemaker, the URT toolkit, the PBM tools, xv, my MC renderer and various
additional scripts and hacks to put everything together. Enjoy!
This file is maintained by Eric Lafortune