usage of rtrace within a material

Dear Lars,

I had the same problem approximately two years ago: rtrace -I does not
give you the right answer when being applied after a dielectric surface.
With the -i option of rtrace, I do not have any experience.
I suggest a workaround for your problem. As you did, I would chose a
dielectric surface followed by a plastic surface. For rtrace -I inside a
material, in theory the integral

E = int [ L cos(theta_m) d Omega_m ]

has to be calculated, m naming the angles inside the material.
Unfortunately, at least the specular path does not remember the light
refraction for the rtrace -I calculation. However, if you substitute the
angles in the integral by the old ones before the material, rtrace -I
works. The substitution theta_m=arcsin(n1/n2*theta), including
cos(theta_m)=sqrt[1-sin^2(theta_m)], leads to a complicated integral that
finally can be simplified to

E = n1^2/n2^2 int [ L cos(theta) d Omega ]

rtrace -I only calculates this integral. With the correction factor
n1^2/n2^2, the rtrace -I results can be applied.
The ambient path, however, seems to work perfectly without any correction
factor.
I haven't done any research on what is exactly going on in the source
codes. However, the results are fine.
I do not have an answer for your third question.

Cheers,
Wendelin

Original message:

···

Date: Thu, 31 Jan 2013 20:58:57 +0800
From: "Lars O. Grobe" <grobe@gmx.net>
To: Radiance general discussion <radiance-general@radiance-online.org>
Subject: [Radiance-general] useage of the dielectric material type
Message-ID: <4CCD6425-B096-44B6-8B39-69741B53E1AD@gmx.net>
Content-Type: text/plain; charset="us-ascii"

Hi all,

while it is one of the most fundamental material types in Radiance,
dielectric is hardly used in simulations with Radiance as far as I know. I
once or twice had to use it, but never for critical parts of a model.

Now, I need to find the irradiance in (!) a glass pane on the second
surface S2 in this simple sketch:

>
>
o | * source
sensor |
>

S2 S1

The glass material applied to a flat surface through S1 obviously does not
work here, as it includes reflection at S2 into the plane through S1. That
would lead to an underestimate of about 4% at normal incidence. What I did
is to model S1 as a dielectric, S2 as plastic, and ran the rtrace command
with the -i switch, location as marked by the o-letter in the sketch and
the view vector towards S2.

- My first question - is this a valid model in Radiance, with a volume
having a dielectric interface on one, a plastic surface on the other side?

- Second question, can I expect rtrace to calculate a valid irradiance
reading under these circumstances?

- Third question, and that is funny - how do I find the transmissivity
parameters for a glass, as dielectric would expect them, if I have
transmission T measured?

If someone here could share some insight how to find a valid dielectric
description from a typical transmission measurement for the visible
spectrum, that would be of great help for me....

Cheers and TIA,
Lars.

Hi Lars,

The physics of this situation is unusual, and the Radiance source code does apply a correction for the "solid angle ratio" caused by lensing at the surface. However, Radiance doesn't track source directions properly inside a dielectric or through a refracting object -- it naively sends rays in a straight line towards the source, missing it in cases where significant refraction occurs. This is why a straight ambient calculation works but the direct calculation does not.

Cheers,
-Greg

···

From: Wendelin.Sprenger@ise.fraunhofer.de
Date: January 31, 2013 8:56:14 AM PST

Dear Lars,

I had the same problem approximately two years ago: rtrace -I does not give you the right answer when being applied after a dielectric surface. With the -i option of rtrace, I do not have any experience.
I suggest a workaround for your problem. As you did, I would chose a dielectric surface followed by a plastic surface. For rtrace -I inside a material, in theory the integral

E = int [ L cos(theta_m) d Omega_m ]

has to be calculated, m naming the angles inside the material. Unfortunately, at least the specular path does not remember the light refraction for the rtrace -I calculation. However, if you substitute the angles in the integral by the old ones before the material, rtrace -I works. The substitution theta_m=arcsin(n1/n2*theta), including cos(theta_m)=sqrt[1-sin^2(theta_m)], leads to a complicated integral that finally can be simplified to

E = n1^2/n2^2 int [ L cos(theta) d Omega ]

rtrace -I only calculates this integral. With the correction factor n1^2/n2^2, the rtrace -I results can be applied.
The ambient path, however, seems to work perfectly without any correction factor.
I haven't done any research on what is exactly going on in the source codes. However, the results are fine.
I do not have an answer for your third question.

Cheers,
Wendelin

Original message:

Date: Thu, 31 Jan 2013 20:58:57 +0800
From: "Lars O. Grobe" <grobe@gmx.net>
To: Radiance general discussion <radiance-general@radiance-online.org>
Subject: [Radiance-general] useage of the dielectric material type
Message-ID: <4CCD6425-B096-44B6-8B39-69741B53E1AD@gmx.net>
Content-Type: text/plain; charset="us-ascii"

Hi all,

while it is one of the most fundamental material types in Radiance, dielectric is hardly used in simulations with Radiance as far as I know. I once or twice had to use it, but never for critical parts of a model.

Now, I need to find the irradiance in (!) a glass pane on the second surface S2 in this simple sketch:

> >
> >
> o | * source
> sensor |
> >

S2 S1

The glass material applied to a flat surface through S1 obviously does not work here, as it includes reflection at S2 into the plane through S1. That would lead to an underestimate of about 4% at normal incidence. What I did is to model S1 as a dielectric, S2 as plastic, and ran the rtrace command with the -i switch, location as marked by the o-letter in the sketch and the view vector towards S2.

- My first question - is this a valid model in Radiance, with a volume having a dielectric interface on one, a plastic surface on the other side?

- Second question, can I expect rtrace to calculate a valid irradiance reading under these circumstances?

- Third question, and that is funny - how do I find the transmissivity parameters for a glass, as dielectric would expect them, if I have transmission T measured?

If someone here could share some insight how to find a valid dielectric description from a typical transmission measurement for the visible spectrum, that would be of great help for me....

Cheers and TIA,
Lars.

I should have said specifically that the source calculation doesn't work if the light bends and stays bent. In a more typical situation where the dielectric is in two parallel planar surfaces, the ray straightens out again and everything works just fine. Radiance started out only having the dielectric material -- glass and trans types were added much later.

Cheers,
-Greg

···

From: Greg Ward <gregoryjward@gmail.com>
Date: January 31, 2013 9:08:04 AM PST

Hi Lars,

The physics of this situation is unusual, and the Radiance source code does apply a correction for the "solid angle ratio" caused by lensing at the surface. However, Radiance doesn't track source directions properly inside a dielectric or through a refracting object -- it naively sends rays in a straight line towards the source, missing it in cases where significant refraction occurs. This is why a straight ambient calculation works but the direct calculation does not.

Cheers,
-Greg

Hi Wendelin, Greg,

thank you both for the helpful replies. It is understandable that the direct calculation was optimized by assuming the direction of the ray before and after transmission to be equal. Especially as thin dielectric volumes were the standard application before the glass modifier came into play. In our days, as we use the dielectric practically only for such exotic cases, one might have decided for a more flexible tracing rather than optimizing, as we have glass. In my current case, I can easily workaround the problem.

Still I am having difficulties now to understand how Wendelin's correction can work if the direct ray would entirely miss the source. Applying a factor to a value of zero should not improve the result. Did you use an extended source of type glow? Or was it rather large, so that (most) rays were hitting it even though the direct calculation sent them towards the wrong direction? It is hard to imagine how this would be possible e.g. with the sun as the only source.

Cheers, Lars.

···

Dear Lars,

I had the same problem approximately two years ago: rtrace -I does not give you the right answer when being applied after a dielectric surface. With the -i option of rtrace, I do not have any experience.
I suggest a workaround for your problem. As you did, I would chose a dielectric surface followed by a plastic surface. For rtrace -I inside a material, in theory the integral

E = int [ L cos(theta_m) d Omega_m ]

has to be calculated, m naming the angles inside the material. Unfortunately, at least the specular path does not remember the light refraction for the rtrace -I calculation. However, if you substitute the angles in the integral by the old ones before the material, rtrace -I works. The substitution theta_m=arcsin(n1/n2*theta), including cos(theta_m)=sqrt[1-sin^2(theta_m)], leads to a complicated integral that finally can be simplified to

E = n1^2/n2^2 int [ L cos(theta) d Omega ]

rtrace -I only calculates this integral. With the correction factor n1^2/n2^2, the rtrace -I results can be applied.
The ambient path, however, seems to work perfectly without any correction factor.
I haven't done any research on what is exactly going on in the source codes. However, the results are fine.
I do not have an answer for your third question.

Cheers,
Wendelin

Original message:

Date: Thu, 31 Jan 2013 20:58:57 +0800
From: "Lars O. Grobe" <grobe@gmx.net>
To: Radiance general discussion <radiance-general@radiance-online.org>
Subject: [Radiance-general] useage of the dielectric material type
Message-ID: <4CCD6425-B096-44B6-8B39-69741B53E1AD@gmx.net>
Content-Type: text/plain; charset="us-ascii"

Hi all,

while it is one of the most fundamental material types in Radiance, dielectric is hardly used in simulations with Radiance as far as I know. I once or twice had to use it, but never for critical parts of a model.

Now, I need to find the irradiance in (!) a glass pane on the second surface S2 in this simple sketch:

> >
> >
> o | * source
> sensor |
> >

S2 S1

The glass material applied to a flat surface through S1 obviously does not work here, as it includes reflection at S2 into the plane through S1. That would lead to an underestimate of about 4% at normal incidence. What I did is to model S1 as a dielectric, S2 as plastic, and ran the rtrace command with the -i switch, location as marked by the o-letter in the sketch and the view vector towards S2.

- My first question - is this a valid model in Radiance, with a volume having a dielectric interface on one, a plastic surface on the other side?

- Second question, can I expect rtrace to calculate a valid irradiance reading under these circumstances?

- Third question, and that is funny - how do I find the transmissivity parameters for a glass, as dielectric would expect them, if I have transmission T measured?

If someone here could share some insight how to find a valid dielectric description from a typical transmission measurement for the visible spectrum, that would be of great help for me....

Cheers and TIA,
Lars.

_______________________________________________
Radiance-general mailing list
Radiance-general@radiance-online.org
http://www.radiance-online.org/mailman/listinfo/radiance-general