genBSDF bug?

Andy told me there was a bug in genBSDF when a BSDF is referenced in
the geometry.

I'm curious if this has been fixed in the current HEAD. Also, can I
test to see if a version of genBSDF is affected by the bug:
1) Run: genBSDF +forward +backward +geom foot
windowAssemblyTest.rad > genBSDFOut.xml
2) Visually compare output from "genBSDFOut.xml" to "70_NoBlinds.xml"

"windowAssemblyTest.rad" contains:
void BSDF testBSDF
6 0 70_NoBlinds.xml 0 1 0 .
0
0

testBSDF polygon testPane
0
0
12
      0 0 0
     30 0 0
     30 30 0
      0 30 0

The base BSDF file is available at
http://www.shuman.org/~tperry/70_NoBlinds.xml It is meant to
represent 70% clear glass.

Thanks for your time.

Tim

Andy and Greg did some tests and confirmed this bug has been fixed.

The oddity I found with the test described below is not related to the
bug. The oddity is simply due to spreading of the distribution during
sampling. This is to be expected with the Klems representation. The
lesson I've learned from this is that you are better off rebuilding
the geometry from scratch rather than using a BSDF with-in a run of
genBSDF. The lack of granularity in the Klems BSDF causes increasing
noise when you iterate through successive Klems BSDF representations.

Good night and good luck.
Tim

···

On Wed, Sep 19, 2012 at 12:05 PM, Tim Perry <[email protected]> wrote:

Andy told me there was a bug in genBSDF when a BSDF is referenced in
the geometry.

I'm curious if this has been fixed in the current HEAD. Also, can I
test to see if a version of genBSDF is affected by the bug:
1) Run: genBSDF +forward +backward +geom foot
windowAssemblyTest.rad > genBSDFOut.xml
2) Visually compare output from "genBSDFOut.xml" to "70_NoBlinds.xml"

"windowAssemblyTest.rad" contains:
void BSDF testBSDF
6 0 70_NoBlinds.xml 0 1 0 .
0
0

testBSDF polygon testPane
0
0
12
      0 0 0
     30 0 0
     30 30 0
      0 30 0

The base BSDF file is available at
http://www.shuman.org/~tperry/70_NoBlinds.xml It is meant to
represent 70% clear glass.

Thanks for your time.

Tim