MPI Informatics Building Model v1.0 - Reported Bugs and Problems

This is the list of problems, bugs, and recommendations concerning the MPI Informatics Building Model v1.0 released on the 12th December 2009. If you want to report a problem, please, first check carefully the list of existing bugs at this webpage. If you have found a new one, write an e-mail with the subject "MPIMODEL V1.0 - new bug report" to Vlastimil Havran. The corrections are planned to another release of dataset. Thank you very much, we appreciate your support and comments.

#1 - Bug - Extra copy of PBRT data (the same content) in the zip file. Corrected to the version v1.0. Reported by Matt Pharr, 5th August 2010.
#2 - Bug - Some materials do not preserve energy (energy conservation law). It will be corrected in the next version 1.05. Reported by Matt Pharr, 5th Auguest 2010.
#3 - Recommendation - PBRT version v2.0 was reported to work correctly for the model as the version 1.06 did crash for higher number of samples per pixel. Reported by Matt Pharr and in PBRT kindly fixed by him as well, 5th August 2010. Thank you very much Matt!
#4 - Recommended setting in PBRT - version v2.0 - for Metropolis renderer. When BRDFs are corrected to be physically plausible (energy conservation law), then the setting for the renderer could be:

Renderer "metropolis"
"float largestepprobability" [0.25]
"integer samplesperpixel" [1024]
"integer bootstrapsamples" [100000]
"integer directsamples" [8]
"bool dodirectseparately" ["true"]
"integer maxconsecutiverejects" [512]
"integer maxdepth" [6]

The contribution by Matt Pharr, reported on 5th August 2010.

This page is maintained by Vlastimil Havran. The usage conditions and copyright for any content on this webpage is given at http://www.mpi-inf.mpg.de/resources/mpimodel/v1.0/index.html#Copyright It was last updated on 12th December 2009.