better results on genBSDF

radiance-general
1

Dear list,

I am using genBSDF to get the BTDF of a shading system, comparing it to an
analytical model using the radiosity method. No surprise, the calculations
seem to be really good, nevertheless, there are some important differences
on directly transmitted (specular) calculations. I am using the following
parameters:

genBSDF -n 4 -c 4000 -r '-ab 4 -ad 512 -as 0 -aa 0 -ds 0.01 -dj 0'
material.mat geometry.rad > ../TMX/file.xml

I tried modifying the -c option, but it did not show any improvements, so I
stayed on 4000. Any suggestions from gurus?

THANKS

German

2

Hi German!

I am using genBSDF to get the BTDF of a shading system, comparing it to
an analytical model using the radiosity method. No surprise, the
calculations seem to be really good, nevertheless, there are some
important differences on directly transmitted (specular) calculations. I

Why do you think that "an analytical model using the radiosity method"
results in "better" results than raytracing does for specular transmission?

Cheers, Lars.

3

Hi Lars,

considering that the specular transmission is that light that does not
touch any surface, the analytical solution would be the exact one (am I
right?). Also, HERE<http://windows.lbl.gov/materials/optics/Bidirectional%20Properties%20of%20Slat%20Shading.pdf&gt;,
those results were compared agains TracePro, and the results of the forward
ray-tracing gave, basically, the same results as the model I am using.

THANKS

German

···

2013/2/16 Lars O. Grobe <[email protected]>

Hi German!

> I am using genBSDF to get the BTDF of a shading system, comparing it to
> an analytical model using the radiosity method. No surprise, the
> calculations seem to be really good, nevertheless, there are some
> important differences on directly transmitted (specular) calculations. I

Why do you think that "an analytical model using the radiosity method"
results in "better" results than raytracing does for specular transmission?

Cheers, Lars.

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

4

The genBSDF program produces a limited resolution matrix representation using the options you've given. This is appropriate if you want to use it with WINDOW (although I think some modifications to the output are still required) or if your material is fairly diffusing. It will not resolve the direct peak to anything finer than 10 degrees, which is the resolution of the full Klems matrix basis. Some spreading of the direct is unavoidable.

You can improve upon this using the tensor tree formulation by setting the -t3 or -t4 option. If your system produces an isotropic distribution (i.e., you can rotate the system about its center with no change to the output), you can try "genBSDF -t3 6" or so. In the more general case, you can use "genBSDF -t4 6", which will resolve the direct component to within a few degrees. You can increase to "-t4 7" to get twice the resolution, but you'll have to increase the -c parameter as well, and I can't predict when the calculation will finish.

The real solution is to incorporate and employ the actual system geometry using the proxy method described near slide 7 in my 2011 workshop presentation:

  http://www.radiance-online.org/community/workshops/2011-berkeley-ca/presentations/day2/GW5_BSDFFirstClass.pdf

Cheers,
-Greg

···

From: Germán Molina Larrain <[email protected]>
Date: February 16, 2013 11:49:09 AM PST

Hi Lars,

considering that the specular transmission is that light that does not touch any surface, the analytical solution would be the exact one (am I right?). Also, HERE, those results were compared agains TracePro, and the results of the forward ray-tracing gave, basically, the same results as the model I am using.

THANKS

German

2013/2/16 Lars O. Grobe <[email protected]>
Hi German!

> I am using genBSDF to get the BTDF of a shading system, comparing it to
> an analytical model using the radiosity method. No surprise, the
> calculations seem to be really good, nevertheless, there are some
> important differences on directly transmitted (specular) calculations. I

Why do you think that "an analytical model using the radiosity method"
results in "better" results than raytracing does for specular transmission?

Cheers, Lars.

---------------------------

From: Germán Molina Larrain <[email protected]>
Date: February 16, 2013 7:06:26 AM PST

Dear list,

I am using genBSDF to get the BTDF of a shading system, comparing it to an analytical model using the radiosity method. No surprise, the calculations seem to be really good, nevertheless, there are some important differences on directly transmitted (specular) calculations. I am using the following parameters:

genBSDF -n 4 -c 4000 -r '-ab 4 -ad 512 -as 0 -aa 0 -ds 0.01 -dj 0' material.mat geometry.rad > ../TMX/file.xml

I tried modifying the -c option, but it did not show any improvements, so I stayed on 4000. Any suggestions from gurus?

THANKS

German

5

Thanks Greg and Lars!

I intend to use these BSDFs for annual simulations (three-phase method). Is
it possible to use the Tensor-tree implementation on that? I think I can
live with the claimed error anyway...

THANKS AGAIN

Germán

···

2013/2/16 Greg Ward <[email protected]>

The genBSDF program produces a limited resolution matrix representation
using the options you've given. This is appropriate if you want to use it
with WINDOW (although I think some modifications to the output are still
required) or if your material is fairly diffusing. It will not resolve the
direct peak to anything finer than 10 degrees, which is the resolution of
the full Klems matrix basis. Some spreading of the direct is unavoidable.

You can improve upon this using the tensor tree formulation by setting the
-t3 or -t4 option. If your system produces an isotropic distribution
(i.e., you can rotate the system about its center with no change to the
output), you can try "genBSDF -t3 6" or so. In the more general case, you
can use "genBSDF -t4 6", which will resolve the direct component to within
a few degrees. You can increase to "-t4 7" to get twice the resolution,
but you'll have to increase the -c parameter as well, and I can't predict
when the calculation will finish.

The real solution is to incorporate and employ the actual system geometry
using the proxy method described near slide 7 in my 2011 workshop
presentation:

http://www.radiance-online.org/community/workshops/2011-berkeley-ca/presentations/day2/GW5_BSDFFirstClass.pdf

Cheers,
-Greg

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 11:49:09 AM PST

*
*

Hi Lars,

considering that the specular transmission is that light that does not
touch any surface, the analytical solution would be the exact one (am I
right?). Also, HERE<http://windows.lbl.gov/materials/optics/Bidirectional%20Properties%20of%20Slat%20Shading.pdf&gt;,
those results were compared agains TracePro, and the results of the forward
ray-tracing gave, basically, the same results as the model I am using.

THANKS

German

2013/2/16 Lars O. Grobe <[email protected]>

Hi German!

> I am using genBSDF to get the BTDF of a shading system, comparing it to
> an analytical model using the radiosity method. No surprise, the
> calculations seem to be really good, nevertheless, there are some
> important differences on directly transmitted (specular) calculations. I

Why do you think that "an analytical model using the radiosity method"
results in "better" results than raytracing does for specular
transmission?

Cheers, Lars.

---------------------------

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 7:06:26 AM PST

*
*

Dear list,

I am using genBSDF to get the BTDF of a shading system, comparing it to an
analytical model using the radiosity method. No surprise, the calculations
seem to be really good, nevertheless, there are some important differences
on directly transmitted (specular) calculations. I am using the following
parameters:

genBSDF -n 4 -c 4000 -r '-ab 4 -ad 512 -as 0 -aa 0 -ds 0.01 -dj 0'
material.mat geometry.rad > ../TMX/file.xml

I tried modifying the -c option, but it did not show any improvements, so
I stayed on 4000. Any suggestions from gurus?

THANKS

German

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

6

The 3-phase method is restricted to matrix (Klems) BSDFs for the time being. A more sophisticated method is in the works, but won't be ready for several months.

Cheers,
-Greg

···

From: Germán Molina Larrain <[email protected]>
Date: February 16, 2013 12:27:02 PM PST

Thanks Greg and Lars!

I intend to use these BSDFs for annual simulations (three-phase method). Is it possible to use the Tensor-tree implementation on that? I think I can live with the claimed error anyway...

THANKS AGAIN

Germán

2013/2/16 Greg Ward <[email protected]>
The genBSDF program produces a limited resolution matrix representation using the options you've given. This is appropriate if you want to use it with WINDOW (although I think some modifications to the output are still required) or if your material is fairly diffusing. It will not resolve the direct peak to anything finer than 10 degrees, which is the resolution of the full Klems matrix basis. Some spreading of the direct is unavoidable.

You can improve upon this using the tensor tree formulation by setting the -t3 or -t4 option. If your system produces an isotropic distribution (i.e., you can rotate the system about its center with no change to the output), you can try "genBSDF -t3 6" or so. In the more general case, you can use "genBSDF -t4 6", which will resolve the direct component to within a few degrees. You can increase to "-t4 7" to get twice the resolution, but you'll have to increase the -c parameter as well, and I can't predict when the calculation will finish.

The real solution is to incorporate and employ the actual system geometry using the proxy method described near slide 7 in my 2011 workshop presentation:

  http://www.radiance-online.org/community/workshops/2011-berkeley-ca/presentations/day2/GW5_BSDFFirstClass.pdf

Cheers,
-Greg

From: Germán Molina Larrain <[email protected]>
Date: February 16, 2013 11:49:09 AM PST

Hi Lars,

considering that the specular transmission is that light that does not touch any surface, the analytical solution would be the exact one (am I right?). Also, HERE, those results were compared agains TracePro, and the results of the forward ray-tracing gave, basically, the same results as the model I am using.

THANKS

German

2013/2/16 Lars O. Grobe <[email protected]>
Hi German!

> I am using genBSDF to get the BTDF of a shading system, comparing it to
> an analytical model using the radiosity method. No surprise, the
> calculations seem to be really good, nevertheless, there are some
> important differences on directly transmitted (specular) calculations. I

Why do you think that "an analytical model using the radiosity method"
results in "better" results than raytracing does for specular transmission?

Cheers, Lars.

---------------------------

From: Germán Molina Larrain <[email protected]>
Date: February 16, 2013 7:06:26 AM PST

Dear list,

I am using genBSDF to get the BTDF of a shading system, comparing it to an analytical model using the radiosity method. No surprise, the calculations seem to be really good, nevertheless, there are some important differences on directly transmitted (specular) calculations. I am using the following parameters:

genBSDF -n 4 -c 4000 -r '-ab 4 -ad 512 -as 0 -aa 0 -ds 0.01 -dj 0' material.mat geometry.rad > ../TMX/file.xml

I tried modifying the -c option, but it did not show any improvements, so I stayed on 4000. Any suggestions from gurus?

THANKS

German

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

7

German,
Did you consider that genBSDF integrates over the incident Klems patch
while radiosity method used in window 6 just uses the center of the patch
for incident energy?
It is easy to modify genBSDF to use the ceter of the Klems patch by
removing a couple of the random variables in incident ray generation.

Andy

···

On Sat, Feb 16, 2013 at 12:49 PM, Greg Ward <[email protected]> wrote:

The 3-phase method is restricted to matrix (Klems) BSDFs for the time
being. A more sophisticated method is in the works, but won't be ready for
several months.

Cheers,
-Greg

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 12:27:02 PM PST

*
*

Thanks Greg and Lars!

I intend to use these BSDFs for annual simulations (three-phase method).
Is it possible to use the Tensor-tree implementation on that? I think I can
live with the claimed error anyway...

THANKS AGAIN

Germán

2013/2/16 Greg Ward <[email protected]>

The genBSDF program produces a limited resolution matrix representation
using the options you've given. This is appropriate if you want to use it
with WINDOW (although I think some modifications to the output are still
required) or if your material is fairly diffusing. It will not resolve the
direct peak to anything finer than 10 degrees, which is the resolution of
the full Klems matrix basis. Some spreading of the direct is unavoidable.

You can improve upon this using the tensor tree formulation by setting
the -t3 or -t4 option. If your system produces an isotropic distribution
(i.e., you can rotate the system about its center with no change to the
output), you can try "genBSDF -t3 6" or so. In the more general case, you
can use "genBSDF -t4 6", which will resolve the direct component to within
a few degrees. You can increase to "-t4 7" to get twice the resolution,
but you'll have to increase the -c parameter as well, and I can't predict
when the calculation will finish.

The real solution is to incorporate and employ the actual system geometry
using the proxy method described near slide 7 in my 2011 workshop
presentation:

http://www.radiance-online.org/community/workshops/2011-berkeley-ca/presentations/day2/GW5_BSDFFirstClass.pdf

Cheers,
-Greg

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 11:49:09 AM PST

*
*

Hi Lars,

considering that the specular transmission is that light that does not
touch any surface, the analytical solution would be the exact one (am I
right?). Also, HERE<http://windows.lbl.gov/materials/optics/Bidirectional%20Properties%20of%20Slat%20Shading.pdf&gt;,
those results were compared agains TracePro, and the results of the forward
ray-tracing gave, basically, the same results as the model I am using.

THANKS

German

2013/2/16 Lars O. Grobe <[email protected]>

Hi German!

> I am using genBSDF to get the BTDF of a shading system, comparing it to
> an analytical model using the radiosity method. No surprise, the
> calculations seem to be really good, nevertheless, there are some
> important differences on directly transmitted (specular) calculations.
I

Why do you think that "an analytical model using the radiosity method"
results in "better" results than raytracing does for specular
transmission?

Cheers, Lars.

---------------------------

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 7:06:26 AM PST

*
*

Dear list,

I am using genBSDF to get the BTDF of a shading system, comparing it to
an analytical model using the radiosity method. No surprise, the
calculations seem to be really good, nevertheless, there are some important
differences on directly transmitted (specular) calculations. I am using the
following parameters:

genBSDF -n 4 -c 4000 -r '-ab 4 -ad 512 -as 0 -aa 0 -ds 0.01 -dj 0'
material.mat geometry.rad > ../TMX/file.xml

I tried modifying the -c option, but it did not show any improvements, so
I stayed on 4000. Any suggestions from gurus?

THANKS

German

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

8

Andy, indeed I did not do that; probably that is what causes the problem...

Could you explain me how to do that?

THANKS

···

2013/2/19 Andrew McNeil <[email protected]>

German,
Did you consider that genBSDF integrates over the incident Klems patch
while radiosity method used in window 6 just uses the center of the patch
for incident energy?
It is easy to modify genBSDF to use the ceter of the Klems patch by
removing a couple of the random variables in incident ray generation.

Andy

On Sat, Feb 16, 2013 at 12:49 PM, Greg Ward <[email protected]>wrote:

The 3-phase method is restricted to matrix (Klems) BSDFs for the time
being. A more sophisticated method is in the works, but won't be ready for
several months.

Cheers,
-Greg

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 12:27:02 PM PST

*
*

Thanks Greg and Lars!

I intend to use these BSDFs for annual simulations (three-phase method).
Is it possible to use the Tensor-tree implementation on that? I think I can
live with the claimed error anyway...

THANKS AGAIN

Germán

2013/2/16 Greg Ward <[email protected]>

The genBSDF program produces a limited resolution matrix representation
using the options you've given. This is appropriate if you want to use it
with WINDOW (although I think some modifications to the output are still
required) or if your material is fairly diffusing. It will not resolve the
direct peak to anything finer than 10 degrees, which is the resolution of
the full Klems matrix basis. Some spreading of the direct is unavoidable.

You can improve upon this using the tensor tree formulation by setting
the -t3 or -t4 option. If your system produces an isotropic distribution
(i.e., you can rotate the system about its center with no change to the
output), you can try "genBSDF -t3 6" or so. In the more general case, you
can use "genBSDF -t4 6", which will resolve the direct component to within
a few degrees. You can increase to "-t4 7" to get twice the resolution,
but you'll have to increase the -c parameter as well, and I can't predict
when the calculation will finish.

The real solution is to incorporate and employ the actual system
geometry using the proxy method described near slide 7 in my 2011 workshop
presentation:

http://www.radiance-online.org/community/workshops/2011-berkeley-ca/presentations/day2/GW5_BSDFFirstClass.pdf

Cheers,
-Greg

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 11:49:09 AM PST

*
*

Hi Lars,

considering that the specular transmission is that light that does not
touch any surface, the analytical solution would be the exact one (am I
right?). Also, HERE<http://windows.lbl.gov/materials/optics/Bidirectional%20Properties%20of%20Slat%20Shading.pdf&gt;,
those results were compared agains TracePro, and the results of the forward
ray-tracing gave, basically, the same results as the model I am using.

THANKS

German

2013/2/16 Lars O. Grobe <[email protected]>

Hi German!

> I am using genBSDF to get the BTDF of a shading system, comparing it
to
> an analytical model using the radiosity method. No surprise, the
> calculations seem to be really good, nevertheless, there are some
> important differences on directly transmitted (specular)
calculations. I

Why do you think that "an analytical model using the radiosity method"
results in "better" results than raytracing does for specular
transmission?

Cheers, Lars.

---------------------------

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 7:06:26 AM PST

*
*

Dear list,

I am using genBSDF to get the BTDF of a shading system, comparing it to
an analytical model using the radiosity method. No surprise, the
calculations seem to be really good, nevertheless, there are some important
differences on directly transmitted (specular) calculations. I am using the
following parameters:

genBSDF -n 4 -c 4000 -r '-ab 4 -ad 512 -as 0 -aa 0 -ds 0.01 -dj 0'
material.mat geometry.rad > ../TMX/file.xml

I tried modifying the -c option, but it did not show any improvements,
so I stayed on 4000. Any suggestions from gurus?

THANKS

German

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

9

actually, now that you mention it, I think that the errors I thought were
caused by the borders can be explained by what you say.

···

2013/2/19 Germán Molina Larrain <[email protected]>

Andy, indeed I did not do that; probably that is what causes the problem...

Could you explain me how to do that?

THANKS

2013/2/19 Andrew McNeil <[email protected]>

German,
Did you consider that genBSDF integrates over the incident Klems patch
while radiosity method used in window 6 just uses the center of the patch
for incident energy?
It is easy to modify genBSDF to use the ceter of the Klems patch by
removing a couple of the random variables in incident ray generation.

Andy

On Sat, Feb 16, 2013 at 12:49 PM, Greg Ward <[email protected]>wrote:

The 3-phase method is restricted to matrix (Klems) BSDFs for the time
being. A more sophisticated method is in the works, but won't be ready for
several months.

Cheers,
-Greg

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 12:27:02 PM PST

*
*

Thanks Greg and Lars!

I intend to use these BSDFs for annual simulations (three-phase method).
Is it possible to use the Tensor-tree implementation on that? I think I can
live with the claimed error anyway...

THANKS AGAIN

Germán

2013/2/16 Greg Ward <[email protected]>

The genBSDF program produces a limited resolution matrix representation
using the options you've given. This is appropriate if you want to use it
with WINDOW (although I think some modifications to the output are still
required) or if your material is fairly diffusing. It will not resolve the
direct peak to anything finer than 10 degrees, which is the resolution of
the full Klems matrix basis. Some spreading of the direct is unavoidable.

You can improve upon this using the tensor tree formulation by setting
the -t3 or -t4 option. If your system produces an isotropic distribution
(i.e., you can rotate the system about its center with no change to the
output), you can try "genBSDF -t3 6" or so. In the more general case, you
can use "genBSDF -t4 6", which will resolve the direct component to within
a few degrees. You can increase to "-t4 7" to get twice the resolution,
but you'll have to increase the -c parameter as well, and I can't predict
when the calculation will finish.

The real solution is to incorporate and employ the actual system
geometry using the proxy method described near slide 7 in my 2011 workshop
presentation:

http://www.radiance-online.org/community/workshops/2011-berkeley-ca/presentations/day2/GW5_BSDFFirstClass.pdf

Cheers,
-Greg

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 11:49:09 AM PST

*
*

Hi Lars,

considering that the specular transmission is that light that does not
touch any surface, the analytical solution would be the exact one (am I
right?). Also, HERE<http://windows.lbl.gov/materials/optics/Bidirectional%20Properties%20of%20Slat%20Shading.pdf&gt;,
those results were compared agains TracePro, and the results of the forward
ray-tracing gave, basically, the same results as the model I am using.

THANKS

German

2013/2/16 Lars O. Grobe <[email protected]>

Hi German!

> I am using genBSDF to get the BTDF of a shading system, comparing it
to
> an analytical model using the radiosity method. No surprise, the
> calculations seem to be really good, nevertheless, there are some
> important differences on directly transmitted (specular)
calculations. I

Why do you think that "an analytical model using the radiosity method"
results in "better" results than raytracing does for specular
transmission?

Cheers, Lars.

---------------------------

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 7:06:26 AM PST

*
*

Dear list,

I am using genBSDF to get the BTDF of a shading system, comparing it to
an analytical model using the radiosity method. No surprise, the
calculations seem to be really good, nevertheless, there are some important
differences on directly transmitted (specular) calculations. I am using the
following parameters:

genBSDF -n 4 -c 4000 -r '-ab 4 -ad 512 -as 0 -aa 0 -ds 0.01 -dj 0'
material.mat geometry.rad > ../TMX/file.xml

I tried modifying the -c option, but it did not show any improvements,
so I stayed on 4000. Any suggestions from gurus?

THANKS

German

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

10

I think all you need to do is change lines 350 and 351 of genBSDF.pl so
they look like this:

Kazi = 360*DEGREE * Kcol / Knaz(Krow);
Kpol = DEGREE * (0.5*Kpola(Krow) + 0.5*Kpola(Krow-1));

(this only changes it for Klems BSDF generation)

Andy

···

On Wed, Feb 20, 2013 at 5:44 AM, Germán Molina Larrain <[email protected]>wrote:

actually, now that you mention it, I think that the errors I thought were
caused by the borders can be explained by what you say.

2013/2/19 Germán Molina Larrain <[email protected]>

Andy, indeed I did not do that; probably that is what causes the
problem...

Could you explain me how to do that?

THANKS

2013/2/19 Andrew McNeil <[email protected]>

German,
Did you consider that genBSDF integrates over the incident Klems patch
while radiosity method used in window 6 just uses the center of the patch
for incident energy?
It is easy to modify genBSDF to use the ceter of the Klems patch by
removing a couple of the random variables in incident ray generation.

Andy

On Sat, Feb 16, 2013 at 12:49 PM, Greg Ward <[email protected]>wrote:

The 3-phase method is restricted to matrix (Klems) BSDFs for the time
being. A more sophisticated method is in the works, but won't be ready for
several months.

Cheers,
-Greg

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 12:27:02 PM PST

*
*

Thanks Greg and Lars!

I intend to use these BSDFs for annual simulations (three-phase
method). Is it possible to use the Tensor-tree implementation on that? I
think I can live with the claimed error anyway...

THANKS AGAIN

Germán

2013/2/16 Greg Ward <[email protected]>

The genBSDF program produces a limited resolution matrix
representation using the options you've given. This is appropriate if you
want to use it with WINDOW (although I think some modifications to the
output are still required) or if your material is fairly diffusing. It
will not resolve the direct peak to anything finer than 10 degrees, which
is the resolution of the full Klems matrix basis. Some spreading of the
direct is unavoidable.

You can improve upon this using the tensor tree formulation by setting
the -t3 or -t4 option. If your system produces an isotropic distribution
(i.e., you can rotate the system about its center with no change to the
output), you can try "genBSDF -t3 6" or so. In the more general case, you
can use "genBSDF -t4 6", which will resolve the direct component to within
a few degrees. You can increase to "-t4 7" to get twice the resolution,
but you'll have to increase the -c parameter as well, and I can't predict
when the calculation will finish.

The real solution is to incorporate and employ the actual system
geometry using the proxy method described near slide 7 in my 2011 workshop
presentation:

http://www.radiance-online.org/community/workshops/2011-berkeley-ca/presentations/day2/GW5_BSDFFirstClass.pdf

Cheers,
-Greg

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 11:49:09 AM PST

*
*

Hi Lars,

considering that the specular transmission is that light that does not
touch any surface, the analytical solution would be the exact one (am I
right?). Also, HERE<http://windows.lbl.gov/materials/optics/Bidirectional%20Properties%20of%20Slat%20Shading.pdf&gt;,
those results were compared agains TracePro, and the results of the forward
ray-tracing gave, basically, the same results as the model I am using.

THANKS

German

2013/2/16 Lars O. Grobe <[email protected]>

Hi German!

> I am using genBSDF to get the BTDF of a shading system, comparing
it to
> an analytical model using the radiosity method. No surprise, the
> calculations seem to be really good, nevertheless, there are some
> important differences on directly transmitted (specular)
calculations. I

Why do you think that "an analytical model using the radiosity method"
results in "better" results than raytracing does for specular
transmission?

Cheers, Lars.

---------------------------

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 7:06:26 AM PST

*
*

Dear list,

I am using genBSDF to get the BTDF of a shading system, comparing it
to an analytical model using the radiosity method. No surprise, the
calculations seem to be really good, nevertheless, there are some important
differences on directly transmitted (specular) calculations. I am using the
following parameters:

genBSDF -n 4 -c 4000 -r '-ab 4 -ad 512 -as 0 -aa 0 -ds 0.01 -dj 0'
material.mat geometry.rad > ../TMX/file.xml

I tried modifying the -c option, but it did not show any improvements,
so I stayed on 4000. Any suggestions from gurus?

THANKS

German

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

11

Ok, I will try those changes, and let you know my results.

THANKS ANDY!

···

2013/2/20 Andrew McNeil <[email protected]>

I think all you need to do is change lines 350 and 351 of genBSDF.pl so
they look like this:

Kazi = 360*DEGREE * Kcol / Knaz(Krow);
Kpol = DEGREE * (0.5*Kpola(Krow) + 0.5*Kpola(Krow-1));

(this only changes it for Klems BSDF generation)

Andy

On Wed, Feb 20, 2013 at 5:44 AM, Germán Molina Larrain <[email protected]>wrote:

actually, now that you mention it, I think that the errors I thought were
caused by the borders can be explained by what you say.

2013/2/19 Germán Molina Larrain <[email protected]>

Andy, indeed I did not do that; probably that is what causes the
problem...

Could you explain me how to do that?

THANKS

2013/2/19 Andrew McNeil <[email protected]>

German,
Did you consider that genBSDF integrates over the incident Klems patch
while radiosity method used in window 6 just uses the center of the patch
for incident energy?
It is easy to modify genBSDF to use the ceter of the Klems patch by
removing a couple of the random variables in incident ray generation.

Andy

On Sat, Feb 16, 2013 at 12:49 PM, Greg Ward <[email protected]>wrote:

The 3-phase method is restricted to matrix (Klems) BSDFs for the time
being. A more sophisticated method is in the works, but won't be ready for
several months.

Cheers,
-Greg

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 12:27:02 PM PST

*
*

Thanks Greg and Lars!

I intend to use these BSDFs for annual simulations (three-phase
method). Is it possible to use the Tensor-tree implementation on that? I
think I can live with the claimed error anyway...

THANKS AGAIN

Germán

2013/2/16 Greg Ward <[email protected]>

The genBSDF program produces a limited resolution matrix
representation using the options you've given. This is appropriate if you
want to use it with WINDOW (although I think some modifications to the
output are still required) or if your material is fairly diffusing. It
will not resolve the direct peak to anything finer than 10 degrees, which
is the resolution of the full Klems matrix basis. Some spreading of the
direct is unavoidable.

You can improve upon this using the tensor tree formulation by
setting the -t3 or -t4 option. If your system produces an isotropic
distribution (i.e., you can rotate the system about its center with no
change to the output), you can try "genBSDF -t3 6" or so. In the more
general case, you can use "genBSDF -t4 6", which will resolve the direct
component to within a few degrees. You can increase to "-t4 7" to get
twice the resolution, but you'll have to increase the -c parameter as well,
and I can't predict when the calculation will finish.

The real solution is to incorporate and employ the actual system
geometry using the proxy method described near slide 7 in my 2011 workshop
presentation:

http://www.radiance-online.org/community/workshops/2011-berkeley-ca/presentations/day2/GW5_BSDFFirstClass.pdf

Cheers,
-Greg

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 11:49:09 AM PST

*
*

Hi Lars,

considering that the specular transmission is that light that does
not touch any surface, the analytical solution would be the exact one (am I
right?). Also, HERE<http://windows.lbl.gov/materials/optics/Bidirectional%20Properties%20of%20Slat%20Shading.pdf&gt;,
those results were compared agains TracePro, and the results of the forward
ray-tracing gave, basically, the same results as the model I am using.

THANKS

German

2013/2/16 Lars O. Grobe <[email protected]>

Hi German!

> I am using genBSDF to get the BTDF of a shading system, comparing
it to
> an analytical model using the radiosity method. No surprise, the
> calculations seem to be really good, nevertheless, there are some
> important differences on directly transmitted (specular)
calculations. I

Why do you think that "an analytical model using the radiosity
method"
results in "better" results than raytracing does for specular
transmission?

Cheers, Lars.

---------------------------

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 7:06:26 AM PST

*
*

Dear list,

I am using genBSDF to get the BTDF of a shading system, comparing it
to an analytical model using the radiosity method. No surprise, the
calculations seem to be really good, nevertheless, there are some important
differences on directly transmitted (specular) calculations. I am using the
following parameters:

genBSDF -n 4 -c 4000 -r '-ab 4 -ad 512 -as 0 -aa 0 -ds 0.01 -dj 0'
material.mat geometry.rad > ../TMX/file.xml

I tried modifying the -c option, but it did not show any
improvements, so I stayed on 4000. Any suggestions from gurus?

THANKS

German

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

12

Andy,

your suggestion worked perfectly.

THANKS

···

2013/2/20 Germán Molina Larrain <[email protected]>

Ok, I will try those changes, and let you know my results.

THANKS ANDY!

2013/2/20 Andrew McNeil <[email protected]>

I think all you need to do is change lines 350 and 351 of genBSDF.pl so
they look like this:

Kazi = 360*DEGREE * Kcol / Knaz(Krow);
Kpol = DEGREE * (0.5*Kpola(Krow) + 0.5*Kpola(Krow-1));

(this only changes it for Klems BSDF generation)

Andy

On Wed, Feb 20, 2013 at 5:44 AM, Germán Molina Larrain <[email protected]>wrote:

actually, now that you mention it, I think that the errors I thought
were caused by the borders can be explained by what you say.

2013/2/19 Germán Molina Larrain <[email protected]>

Andy, indeed I did not do that; probably that is what causes the
problem...

Could you explain me how to do that?

THANKS

2013/2/19 Andrew McNeil <[email protected]>

German,
Did you consider that genBSDF integrates over the incident Klems patch
while radiosity method used in window 6 just uses the center of the patch
for incident energy?
It is easy to modify genBSDF to use the ceter of the Klems patch by
removing a couple of the random variables in incident ray generation.

Andy

On Sat, Feb 16, 2013 at 12:49 PM, Greg Ward <[email protected]>wrote:

The 3-phase method is restricted to matrix (Klems) BSDFs for the time
being. A more sophisticated method is in the works, but won't be ready for
several months.

Cheers,
-Greg

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 12:27:02 PM PST

*
*

Thanks Greg and Lars!

I intend to use these BSDFs for annual simulations (three-phase
method). Is it possible to use the Tensor-tree implementation on that? I
think I can live with the claimed error anyway...

THANKS AGAIN

Germán

2013/2/16 Greg Ward <[email protected]>

The genBSDF program produces a limited resolution matrix
representation using the options you've given. This is appropriate if you
want to use it with WINDOW (although I think some modifications to the
output are still required) or if your material is fairly diffusing. It
will not resolve the direct peak to anything finer than 10 degrees, which
is the resolution of the full Klems matrix basis. Some spreading of the
direct is unavoidable.

You can improve upon this using the tensor tree formulation by
setting the -t3 or -t4 option. If your system produces an isotropic
distribution (i.e., you can rotate the system about its center with no
change to the output), you can try "genBSDF -t3 6" or so. In the more
general case, you can use "genBSDF -t4 6", which will resolve the direct
component to within a few degrees. You can increase to "-t4 7" to get
twice the resolution, but you'll have to increase the -c parameter as well,
and I can't predict when the calculation will finish.

The real solution is to incorporate and employ the actual system
geometry using the proxy method described near slide 7 in my 2011 workshop
presentation:

http://www.radiance-online.org/community/workshops/2011-berkeley-ca/presentations/day2/GW5_BSDFFirstClass.pdf

Cheers,
-Greg

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 11:49:09 AM PST

*
*

Hi Lars,

considering that the specular transmission is that light that does
not touch any surface, the analytical solution would be the exact one (am I
right?). Also, HERE<http://windows.lbl.gov/materials/optics/Bidirectional%20Properties%20of%20Slat%20Shading.pdf&gt;,
those results were compared agains TracePro, and the results of the forward
ray-tracing gave, basically, the same results as the model I am using.

THANKS

German

2013/2/16 Lars O. Grobe <[email protected]>

Hi German!

> I am using genBSDF to get the BTDF of a shading system, comparing
it to
> an analytical model using the radiosity method. No surprise, the
> calculations seem to be really good, nevertheless, there are some
> important differences on directly transmitted (specular)
calculations. I

Why do you think that "an analytical model using the radiosity
method"
results in "better" results than raytracing does for specular
transmission?

Cheers, Lars.

---------------------------

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 7:06:26 AM PST

*
*

Dear list,

I am using genBSDF to get the BTDF of a shading system, comparing it
to an analytical model using the radiosity method. No surprise, the
calculations seem to be really good, nevertheless, there are some important
differences on directly transmitted (specular) calculations. I am using the
following parameters:

genBSDF -n 4 -c 4000 -r '-ab 4 -ad 512 -as 0 -aa 0 -ds 0.01 -dj 0'
material.mat geometry.rad > ../TMX/file.xml

I tried modifying the -c option, but it did not show any
improvements, so I stayed on 4000. Any suggestions from gurus?

THANKS

German

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

13

Great! Glad it all works out.

···

On Thu, Feb 21, 2013 at 10:48 AM, Germán Molina Larrain <[email protected]>wrote:

Andy,

your suggestion worked perfectly.

THANKS

2013/2/20 Germán Molina Larrain <[email protected]>

Ok, I will try those changes, and let you know my results.

THANKS ANDY!

2013/2/20 Andrew McNeil <[email protected]>

I think all you need to do is change lines 350 and 351 of genBSDF.pl so
they look like this:

Kazi = 360*DEGREE * Kcol / Knaz(Krow);
Kpol = DEGREE * (0.5*Kpola(Krow) + 0.5*Kpola(Krow-1));

(this only changes it for Klems BSDF generation)

Andy

On Wed, Feb 20, 2013 at 5:44 AM, Germán Molina Larrain <[email protected]>wrote:

actually, now that you mention it, I think that the errors I thought
were caused by the borders can be explained by what you say.

2013/2/19 Germán Molina Larrain <[email protected]>

Andy, indeed I did not do that; probably that is what causes the
problem...

Could you explain me how to do that?

THANKS

2013/2/19 Andrew McNeil <[email protected]>

German,
Did you consider that genBSDF integrates over the incident Klems
patch while radiosity method used in window 6 just uses the center of the
patch for incident energy?
It is easy to modify genBSDF to use the ceter of the Klems patch by
removing a couple of the random variables in incident ray generation.

Andy

On Sat, Feb 16, 2013 at 12:49 PM, Greg Ward <[email protected]>wrote:

The 3-phase method is restricted to matrix (Klems) BSDFs for the
time being. A more sophisticated method is in the works, but won't be
ready for several months.

Cheers,
-Greg

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 12:27:02 PM PST

*
*

Thanks Greg and Lars!

I intend to use these BSDFs for annual simulations (three-phase
method). Is it possible to use the Tensor-tree implementation on that? I
think I can live with the claimed error anyway...

THANKS AGAIN

Germán

2013/2/16 Greg Ward <[email protected]>

The genBSDF program produces a limited resolution matrix
representation using the options you've given. This is appropriate if you
want to use it with WINDOW (although I think some modifications to the
output are still required) or if your material is fairly diffusing. It
will not resolve the direct peak to anything finer than 10 degrees, which
is the resolution of the full Klems matrix basis. Some spreading of the
direct is unavoidable.

You can improve upon this using the tensor tree formulation by
setting the -t3 or -t4 option. If your system produces an isotropic
distribution (i.e., you can rotate the system about its center with no
change to the output), you can try "genBSDF -t3 6" or so. In the more
general case, you can use "genBSDF -t4 6", which will resolve the direct
component to within a few degrees. You can increase to "-t4 7" to get
twice the resolution, but you'll have to increase the -c parameter as well,
and I can't predict when the calculation will finish.

The real solution is to incorporate and employ the actual system
geometry using the proxy method described near slide 7 in my 2011 workshop
presentation:

http://www.radiance-online.org/community/workshops/2011-berkeley-ca/presentations/day2/GW5_BSDFFirstClass.pdf

Cheers,
-Greg

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 11:49:09 AM PST

*
*

Hi Lars,

considering that the specular transmission is that light that does
not touch any surface, the analytical solution would be the exact one (am I
right?). Also, HERE<http://windows.lbl.gov/materials/optics/Bidirectional%20Properties%20of%20Slat%20Shading.pdf&gt;,
those results were compared agains TracePro, and the results of the forward
ray-tracing gave, basically, the same results as the model I am using.

THANKS

German

2013/2/16 Lars O. Grobe <[email protected]>

Hi German!

> I am using genBSDF to get the BTDF of a shading system,
comparing it to
> an analytical model using the radiosity method. No surprise, the
> calculations seem to be really good, nevertheless, there are some
> important differences on directly transmitted (specular)
calculations. I

Why do you think that "an analytical model using the radiosity
method"
results in "better" results than raytracing does for specular
transmission?

Cheers, Lars.

---------------------------

*From: *Germán Molina Larrain <[email protected]>

*Date: *February 16, 2013 7:06:26 AM PST

*
*

Dear list,

I am using genBSDF to get the BTDF of a shading system, comparing
it to an analytical model using the radiosity method. No surprise, the
calculations seem to be really good, nevertheless, there are some important
differences on directly transmitted (specular) calculations. I am using the
following parameters:

genBSDF -n 4 -c 4000 -r '-ab 4 -ad 512 -as 0 -aa 0 -ds 0.01 -dj 0'
material.mat geometry.rad > ../TMX/file.xml

I tried modifying the -c option, but it did not show any
improvements, so I stayed on 4000. Any suggestions from gurus?

THANKS

German

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general