Model glass stack with measured reflection

Discussions about LuxRender's materials, and how to make the most of them.

Moderators: Dade, coordinators

Re: Model glass stack with measured reflection

Postby rhRunner » Thu Feb 09, 2017 4:32 am

Hi Jeanphi,

Thank you for your explanation. I tried to illustrate my question in the following figure:
(1)As I need to model the transmittance of the glass with volume, I modeled the 5 faces with glass2 with volume defined with tabulated transmission data.
(2)My question is about the back side of the glass. It has unique reflection property because of the coating. How should I model this polygon?

If I model it with glass2 and volume, I could not define its unique reflection property.
If I model it with glass and volume, then the IOR defined in glass material is conflicted with the IOR defined in volume. I remember when I use glass material and volume, Luxrender will use IOR defined by glass and ignore iOR define in volume.
if I model it with glass without volume, then the glass volume is not sealed.
I also have problem to use fresnel color for glass IOR

Any suggestions? Shall I use layered material?
Capture.PNG
rhRunner
 
Posts: 47
Joined: Thu Sep 18, 2014 12:10 pm

Re: Model glass stack with measured reflection

Postby jeanphi » Thu Feb 09, 2017 6:54 am

Hi,

Why can't you use glass2 to model the back face? Its reflection properties should be able to be represented by the IOR, that's actually what fresnel color texture is about.
If you use the glass material, the volume IOR is completely ignored, you can only use the volume to define the absorption.
The volumes don't need to be explicitely sealed, the volume information is reset at boundaries according to what has been configured. If some volume information is missing, it is inferred from the ray other vertex information, and if that information is also missing then a default volume is used.
What is your issue with fresnel color? Fresnel color can create an IOR with a complex part that will cause additional absorption, is that the issue?
I don't think layered material will work much better, but you can try it.

Jeanphi
jeanphi
Developer
 
Posts: 7943
Joined: Mon Jan 14, 2008 7:21 am

Re: Model glass stack with measured reflection

Postby jantschek » Thu Feb 09, 2017 7:13 am

Hi,

I've set an *.blend like it is attached. Unfortunatly you're going to reset the file references.
* upcoming point * how do I create such an pack and go file for blender?

Anyway I did my best do exclude the necessary input data from the tables... I hope they're right but nobody is perfect.
This my result using that properties. I'm not sure about the rainbow effect - no idea where it's coming from...
But the multireflection is very well visible... (be carefull about the size of the object - it must be tiny compares to the thickness of the glass otherwise the reflections can't be separated)

Guardian.png


BR
Attachments
Guadian.zip
(1.48 MiB) Downloaded 36 times
Lux Render Version: 1.7 (in classic mode)
Exporter: Blender
Exporter Version: 2.79
Operating System: Windows 7 X64
jantschek
 
Posts: 187
Joined: Thu Jul 09, 2015 11:02 am

Re: Model glass stack with measured reflection

Postby rhRunner » Thu Feb 09, 2017 11:08 am

Hi Jeanphi,

I am not sure whether I mentioned this, the measured reflection data for the glass is mostly coming from the coating, not from the glass/air interface. The coating is at face 2 where the dashed line is. I used glass 2 for entire glass and use tabulated IOR. As shown in the above image, if light come from left, then reflection happens at the first glass/air interface, not at the coating (dashed line). Does this make sense?

In Jantschek's solution, he use two volume definitions:
(1)"clear glass volume", IOR defined by clear glass reflection tabulated data, absorption defined by clear glass transmission tabulated data
(2)"coating volume", IOR defined by coated glass reflection tabulated data, absorption defined by clear glass transmission tabulated data

The five faces (front side of the glass and edges) is defined as glass2 with "clear glass volume". Back face defined as glass2 with "coating volume". Is this going to cause confusion? Light enter glass volume from left assuming IOR in "clear glass volume". When it hit the back side of the glass, IOR inside the volume is updated to "coating volume". Is this how volume works?

For fresnel color:

I have reflection data in a file:

380 0.5
385 0.6
.
.
.
I use this file in tabulated texture and then use this to texture fresnel color. Is this how it works?
This works well in volume IOR but when I use the same method on glass IOR, luxrender complains that it cannot find the fresnel color texture.
rhRunner
 
Posts: 47
Joined: Thu Sep 18, 2014 12:10 pm

Re: Model glass stack with measured reflection

Postby rhRunner » Thu Feb 09, 2017 11:29 am

Hi Jantschek,
The result is absolute amazing and thank you so much for sharing the blender file.
I am trying to learn how you defines luxpop. I noticed that absorption coefficient in excel file is calculated from transmission data. Should this be calculated from (1-transmission-front reflection)?
I am also a bit confuse by the way you calculate IOR. What is unified reflectivity and how does it lead to IOR? Is it the same if I use fresnel color with reflection data?
rhRunner
 
Posts: 47
Joined: Thu Sep 18, 2014 12:10 pm

Re: Model glass stack with measured reflection

Postby jantschek » Thu Feb 09, 2017 2:57 pm

Hi,

EDIT
rhRunner you're right. I messed up the calculation for absorption coefficient.
Really sorry... so this image is wrong... I'll fix it tomorrow.


I read wikipedia (once again) for absorption.
https://de.wikipedia.org/wiki/Absorptionskoeffizient
Guardian_ABS_coefficient.png

Sorry for confusing you but using the transmission for absorption calculation is right.

If you use the pure transmission or the transmission+reflectance depends on the measurement method that we don't know exactly.

Since its called ultra clear or even extra clear I'll assume that I'll use the sum of transmission and reflection to correct my mistake. So the amount of light that is not transmitted and not reflected is 1-(T×2R)... the result should be some very little values for visible light. We'll see...

EDIT
The remaining question is whether the measured transmission in the datasheet has been corrected by the reflection already or not.
A deeper look into the data shows that some value pairs generate values larger than 1 if you sum up the transmission and reflectance values.
The result would be a negative absorption that is not possible afaik.
Thats the main reason for me to choose the pure transmission value for absorption calculation.
On the other hand this could be due to measurement variations and we could simply set all negative coefficients to zero and compare both rendering images.

For IOR calculation i have used the fresnel formula for perpendicular incidence angles. You can check it on Wikipedia.
The unified is only a nearing. Theoretical front and back reflection is equal on non coated materials. So I simply took the average.

EDIT
Shame on me
(Shame on me)²
BR
Lux Render Version: 1.7 (in classic mode)
Exporter: Blender
Exporter Version: 2.79
Operating System: Windows 7 X64
jantschek
 
Posts: 187
Joined: Thu Jul 09, 2015 11:02 am

Re: Model glass stack with measured reflection

Postby rhRunner » Fri Feb 10, 2017 10:34 am

Hi Jantschek,

Thank you for your explanation. There are standard for measuring reflection and transmission of glazing. It normally involved an integrate sphere. http://www.azom.com/images/Article_Images/ImageForArticle_13488(4).jpg
I think the transmittance value is the ratio between transmitted/total. I checked the file and I did not find any value where
front reflection+transmission > 1. Where does the total value larger than 1 happen to you?

For IOR calculation, I am still confused how you did the calculation. You have
(-(reflection+2*SQRT(reflection)+1))/(reflection-1).

From Wikipedia and jeanphi's earlier suggestion, should this be (SQRT(reflection)+1)/(SQRT(reflection)+1)? Did I miss anything?

I also realized that changing IOR to adapt to reflection data might not be a good idea for this glazing sample. As the name "SunGuard Silver " suggested, the coating mixed silver in it which causes the mirror like reflection. Probably layered material is better in this case. I tried the IOR and it bends light so much and cannot see through the sample anymore.
test_IOR_Coating.png
rhRunner
 
Posts: 47
Joined: Thu Sep 18, 2014 12:10 pm

Re: Model glass stack with measured reflection

Postby jantschek » Fri Feb 10, 2017 3:37 pm

Hi,

thanks. I hope the measurement is not done using an integration sphere. Otherwise it will become hard determine the IOR, because this simple frensel formula is valid only for perpendicular incidence.
EDIT: Except there is a collimated light source in use the ensures perpendicular incidence. Do you have some more information regarding that?

Usually reflections happens on both interfaces. For glases typically something about 4% from air into glass and another 4% from glas back into air.

I think the transmittance value is the ratio between transmitted/total. I checked the file and I did not find any value where front reflection+transmission > 1. Where does the total value larger than 1 happen to you?


So the sum of both reflections + the transmission can't be larger than 1.

From Wikipedia and jeanphi's earlier suggestion, should this be (SQRT(reflection)+1)/(SQRT(reflection)+1)? Did I miss anything?


I hope you meant (SQRT(reflection)+1)/(1-SQRT(reflection)). This formula is exactly the same like mine but in a better way. You forced me to do the math, thanks a lot :D
My result came from the calculator, yours look much better.

I tried the IOR and it bends light so much and cannot see through the sample anymore.

I had the same issue - I think I have to rework on the setup. It might work with a new thin geometric layer doing another interface coating to glass. I'll give it a try as soon as I'll find some time.
That means also that we have to rework the properties, but at the moment I don't have any idea how that should be done.

Probably layered material is better in this case

I have a feeling that it doesn't work regarding physical correctness...

BR
Lux Render Version: 1.7 (in classic mode)
Exporter: Blender
Exporter Version: 2.79
Operating System: Windows 7 X64
jantschek
 
Posts: 187
Joined: Thu Jul 09, 2015 11:02 am

Re: Model glass stack with measured reflection

Postby jantschek » Thu May 11, 2017 10:48 pm

Hi,

it took me some time, but I've started some more investigations on the coated windows.
What do you think about this results?

To see whether there are strange colour effects I did some renderings using a black/white HDRI ambient source.
BW_Silver_Coated_Side.jpg

BW_Clima_Coated_Side.jpg


And some coloured results as well.
Col_Silver_Coated_Side.jpg

Col_Clima_Coated_Side.jpg


No colour effects anymore - and no strange refraction.
From that point I'm quite happy with the results.
What do you think?

There is the scene file including the "new" input files for materials.
Guardian.zip
Sample_Scene
(83.81 KiB) Downloaded 23 times


BR
Lux Render Version: 1.7 (in classic mode)
Exporter: Blender
Exporter Version: 2.79
Operating System: Windows 7 X64
jantschek
 
Posts: 187
Joined: Thu Jul 09, 2015 11:02 am

Re: Model glass stack with measured reflection

Postby rhRunner » Thu Oct 26, 2017 10:35 am

Hi Jantschek,

Thank you very much for the updates. Very sorry for the late reply. The images looks amazing. Actually I did some testing as well. My current understanding is this: for glass with coating (most of them silver or some other metal based), reflection even at normal angle is much higher than IOR=1.5 could generate. If I change IOR to higher value to match reflection at normal angle then there is total reflection beyond certain angle. I end up mixing metal with glass at one surface of glass volume. I attached one image of the glass buildup, multiple reflection is very interesting and it does happen to the sample I have. I tested reflectance and transmittance and both could match measured data well. I haven't check your blender file but the renders looks amazing. Very excited to learn from it.

Xiaoming
Attachments
MultiLayer.png
rhRunner
 
Posts: 47
Joined: Thu Sep 18, 2014 12:10 pm

Previous

Return to Materials Help

Who is online

Users browsing this forum: No registered users and 2 guests