Tensor tree BSDF

Hi list,
The following glass consists of plenty of glazing strips and solar cells and the window is made up of such eight glass.

I wanna generate a high resoluton tensor tree BSDF with genBSDF and the command is like this:

genBSDF -n 24 -t4 6 -c 40960 -geom meter +backward +forward fullwindow.rad

The edge conditions like the frame and the south wall is included in the file fullwindow.rad:

But when this file is used in the sun coefficients part of 5 phase method, the rendered image of one of the glass is all black.

oconv -f materialBlack.rad roomBlack.rad suns.rad glazingbsdf.rad > octrees/sun.oct

The glazingbsdf.rad is as follows:

However, when I used the klems BSDF from WINDOW, the image is like this:
It seems that the problem is related to tensor tree BSDF which maybe not generated accurately.
Any guidance or suggestions are much appreciated!

I am guessing, but since your material has a pure “through” or “vision” component, you had best use the “aBSDF” material type, rather than “BSDF”. This will permit light to pass through the material at the appropriate angles, generating a more accurate sun patch. The only reason you see a sun patch using a Klems file is because the lower BSDF resolution allows some ray samples to hit the sun. There are probably a few such samples in the tensor tree rendering as well if you look for them, but the chances with a high-resolution BSDF are much reduced. The “aBSDF” type is designed to resolve this issue.


Hi @Greg_Ward,
I didn’t find the material aBSDF from the referce document refman. So how to generate this material aBSDF and apply it? Is there any document which we can refer to?

Ah, sorry about that! I see that the file is out of date on the website. We shall update it.

Meanwhile, you can use the same arguments for aBSDF as for BSDF, less the “thickness” parameter, since the aBSDF type does not support proxy geometry. Just remove that first string argument and shift the others over for aBSDF.


Hi @Greg_Ward,
I am not sure if I have understood the usage of aBSDF and the following is the file glazingbsdf.rad that can be used in 5 phase method. And should I use tensor tree BSDF or Klems BSDF?

oconv materialBlack.rad roomBlack.rad suns.rad glazingbsdf.rad > sun.oct

Hi @lee,

You forgot to remove the “thickness” parameter (the first argument). I don’t know if there is any better official documentation than the source code: https://github.com/LBNL-ETA/Radiance/blob/master/src/rt/m_bsdf.c#L58-L63

It should be:

void aBSDF glazing
5 /home/dell/Cities/validation_exp/facing_window/matrices/tmtx/STPV.xml 0 0 1 .
1 Like

Yes, Mikkel is correct, and you can use either a Klems or a tensor tree representation with aBSDF. You might want to try both to see the differences.

The new reference manual is now up, by the way. (Be sure to shift-reload to refresh any local caching.)


Hi @mikkelkp, @Greg_Ward
Thanks for your reply! I am going to correct the file and try these two types of angular basis with aBSDF and compare the differences.