[Radiance-general] Glare analysis of BSDF representation of light redirecting glass

[Greg Ward]

Hi Chris,

Changing the -c setting in genBSDF may result in a more accurate BSDF estimate, but it will not increase the resolution of the recorded distribution, which is defined by the 145x145 Klems basis.  Near normal incidence, the resolution is about 5 degrees, but increases to 15 or 20 degrees towards grazing angles.

To obtain a higher-resolution BSDF, you will need to switch to the tensor tree representation, which means using the -t4 option, probably '-t4 5' or '-t4 6' in your case.  The '-t4 5' setting will yield a BSDF with 1024 incoming and 1024 outgoing directions, which corresponds roughly to 2-degree resolution near normal incidence.  The higher setting of '-t4 6' gives you about twice the resolution, or 1-degree accuracy.  However, it does take about 4 times as long to calculate, and any tensor tree BSDF will take much longer to compute with genBSDF than the Klems basis.  Final rendering times should be similar, however.

Hope this helps!
-Greg

> From: Humann Chris <chris at coolshadow.com>
> Date: January 14, 2013 11:26:16 AM PST
> 
> Dear group,
> 
> I have a BSDF description of a light redirecting glass generated with the following command:
> genBSDF -n 8 -c 20000 -r -ab 6 -ad 1 +f +b -g- millimeter -dim 12.49 12.51 12.47449 12.52551 -3.158 0 
> 
> We are using this file to understand the effective light distribution of a particular product as well as calculating DGP.   We are wondering if the number of samples (20000) used per each of the 145 klems directions would be sufficient to give an accurate representation of glare when looking at the scene description from inside the space?  I've looked for published studies of comparative glare analysis using DGP and BSDF files but have come up short.
> 
> Thanks as always,
> Chris