The question is on the design of a roof in an exhibition hall. The first layer is facing the outside air, the next layer is approx. 20cm below and the third layer is another few meters below. All 3 layers are built up the diffuse transmitting glass.
We have built all 3 layers of glass with material Trans, cf. the data below:
My question is when rendering the scene and looking up towards the roof you can see right through the panes as they are not translucent but clear glass. One can clearly see outline of the top window and connection between the roof and the walls which is rather unexpected.
Are there a logical explanation for this or what could be the case here? Should I have modeled the inner layer as an illum source?
You should not be seeing straight through the surface. I did a test using your parameters, and could not see through such a trans material.
Are you sure that what you are seeing is not the shadows cast by the beams? Have you double-checked your geometry to make sure all the materials are correctly assigned and the surfaces where you expect them? What happens when you substitute an opaque material for the trans?
-Greg
···
From: Per Haugaard <[email protected]>
Date: January 21, 2013 11:57:34 AM PST
Dear list,
Yet another question on the trans material.
The question is on the design of a roof in an exhibition hall. The first layer is facing the outside air, the next layer is approx. 20cm below and the third layer is another few meters below. All 3 layers are built up the diffuse transmitting glass.
We have built all 3 layers of glass with material Trans, cf. the data below:
My question is when rendering the scene and looking up towards the roof you can see right through the panes as they are not translucent but clear glass. One can clearly see outline of the top window and connection between the roof and the walls which is rather unexpected.
Are there a logical explanation for this or what could be the case here? Should I have modeled the inner layer as an illum source?
You should not be seeing straight through the surface. I did a test using your parameters, and could not see through such a trans material.
Are you sure that what you are seeing is not the shadows cast by the beams? Have you double-checked your geometry to make sure all the materials are correctly assigned and the surfaces where you expect them? What happens when you substitute an opaque material for the trans?
-Greg
From: Per Haugaard <[email protected]>
Date: January 21, 2013 11:57:34 AM PST
Dear list,
Yet another question on the trans material.
The question is on the design of a roof in an exhibition hall. The first layer is facing the outside air, the next layer is approx. 20cm below and the third layer is another few meters below. All 3 layers are built up the diffuse transmitting glass.
We have built all 3 layers of glass with material Trans, cf. the data below:
My question is when rendering the scene and looking up towards the roof you can see right through the panes as they are not translucent but clear glass. One can clearly see outline of the top window and connection between the roof and the walls which is rather unexpected.
Are there a logical explanation for this or what could be the case here? Should I have modeled the inner layer as an illum source?
a side note on your material definition. Ts of zero is difficult to achieve for transmission with real glass, as that would be a perfect lambertian diffuser. Did you do any test or measurement confirming this? You can see some examples of real world translucent glasses (sanded, or with scattering films applied) on Peter's BME site: BME BSDF, BRDF data and models
A quick test I tend to do is pointing a small laser pointer at the glass (be careful, there may be strong reflection from the glass which you probably do not want to reach your eye) held just one cm onver a piece of paper. If you see a smooth gradient, it is a perfect diffuser. If you still see the dot of the pointer, with a smooth gradient around, you have a mixture of diffuse and specular transmission. That ratio could get measured using an intergrating sphere, or you could have the BSDF measured.
The question regarding the Trans material previously posted by Per was
basically posted on my behalf and therefore I will take the liberty to just
build upon that thread.
With respect to the difference between generating illuminance and luminance
pictures you of course have a very valid point. However, I am simulating a
model with three layers of Trans material and I would expect that your view
from inside and outwards even when considering illuminance levels would be
diffused (being filtered through three layers) ? Maybe I am missing
something?
What happens in between layers of Trans material (inter-reflections
etc.) and what are the internal surface properties?
First, it would help to know whether you experience this with luminance maps, too.
For illluminance/irradiance any transmissive surface is tricky, as the observer may expect different results.
What -ab setting do you use? -ab <3 would only let the specular transmission trough your multilayer diffuser, if I am not completely wrong... And that should make it appear like a dark (+ reflection) clear glass? I never tried this kind of stacking.
Cheers, Lars.
···
Hi all,
Hope I am getting how this works right...
The question regarding the Trans material previously posted by Per was basically posted on my behalf and therefore I will take the liberty to just build upon that thread.
With respect to the difference between generating illuminance and luminance pictures you of course have a very valid point. However, I am simulating a model with three layers of Trans material and I would expect that your view from inside and outwards even when considering illuminance levels would be diffused (being filtered through three layers) ? Maybe I am missing something?
What happens in between layers of Trans material (inter-reflections etc.) and what are the internal surface properties?
Any trans materials will be ignored if you use the -i rendering option. If you didn't use -i+, then you shouldn't see through a trans with Ts=0.
Between layers of trans materials with both diffuse reflection and transmission, many interreflected rays will be traced.
-Greg
···
From: Martin Vraa Nielsen <[email protected]>
Date: January 22, 2013 6:03:29 AM PST
Hi all,
Hope I am getting how this works right...
The question regarding the Trans material previously posted by Per was basically posted on my behalf and therefore I will take the liberty to just build upon that thread.
With respect to the difference between generating illuminance and luminance pictures you of course have a very valid point. However, I am simulating a model with three layers of Trans material and I would expect that your view from inside and outwards even when considering illuminance levels would be diffused (being filtered through three layers) ? Maybe I am missing something?
What happens in between layers of Trans material (inter-reflections etc.) and what are the internal surface properties?
Thank you all for your help - is was very beneficial.
You were right that it appears to be "perfectly" diffused when considering
luminance. However, just for future references, isn't it possible to
evaluate the illuminance incident on a surface with trans or glass material
applied?
Greg: with respect to the -i rendering option or not, I must admit I simply
do not know enough to have a sufficient grasp on this. As I read it from
the manual -i basically determined if you consider radiance or irradiance
and therefore subsequently if you consider illuminance or luminance. Do I
understand that correctly?
Sorry for all the potential trivial and basic questions and thank you for
taking the time to answer them!
You were right that it appears to be "perfectly" diffused when considering luminance. However, just for future references, isn't it possible to evaluate the illuminance incident on a surface with trans or glass material applied?
You can do so by piping "sensor locations/directions" into rtrace and use the -I+ switch. That way, you calculate irradiance at that location without having a visible surface in place. You can get one value for a given location by e.g. piping one location / direction vector, e.g. echo "0 0 0 0 0 1" | rtrace -I+ (all other parameters...). Or you can generate a grid of such values, eventuelly forming an "irradiance image", by using vwrays to generate the locations / directions for you. This is much more flexible then rpict -i, which always depends on a surface in your scene. Have a quick look at Axel Jacob's tutorial to see some examples.
Your understanding of the -i option is correct in the basics, but there is a "hack" in the code to ignore glass and other typically transparent surfaces. Otherwise, you would end up calculating illuminance on surfaces that generally don't care about illuminance. And in the case of your diffuse trans material, it is unclear whether you would care more about illuminance on the inside (which would be the default reporting for interior views), or illuminance arriving at the other side. Do you see the problem?
Lars' solution of using rtrace to compute the location then a second rtrace to compute irradiance (or illuminance) using -I+ is the best, because it gives you full control over what gets reported. John Mardaljevic has used this to good effect in several specialized cases, which he has described here:
If you want to speed up the second calculation on a multiprocessing machine, add a -n option to the second rtrace with the number of parallel processes. Be sure to specify an ambient file with -n if -ab > 0, or you won't see much speedup.
Cheers,
-Greg
···
From: "Lars O. Grobe" <[email protected]>
Date: January 25, 2013 2:58:04 AM PST
Hi Martine!
You were right that it appears to be "perfectly" diffused when considering luminance. However, just for future references, isn't it possible to evaluate the illuminance incident on a surface with trans or glass material applied?
You can do so by piping "sensor locations/directions" into rtrace and use the -I+ switch. That way, you calculate irradiance at that location without having a visible surface in place. You can get one value for a given location by e.g. piping one location / direction vector, e.g. echo "0 0 0 0 0 1" | rtrace -I+ (all other parameters...). Or you can generate a grid of such values, eventuelly forming an "irradiance image", by using vwrays to generate the locations / directions for you. This is much more flexible then rpict -i, which always depends on a surface in your scene. Have a quick look at Axel Jacob's tutorial to see some examples.
Cheers, Lars.
----------------
From: Martin Vraa Nielsen <[email protected]>
Date: January 25, 2013 1:49:49 AM PST
Hi all,
Thank you all for your help - is was very beneficial.
You were right that it appears to be "perfectly" diffused when considering luminance. However, just for future references, isn't it possible to evaluate the illuminance incident on a surface with trans or glass material applied?
Greg: with respect to the -i rendering option or not, I must admit I simply do not know enough to have a sufficient grasp on this. As I read it from the manual -i basically determined if you consider radiance or irradiance and therefore subsequently if you consider illuminance or luminance. Do I understand that correctly?
Sorry for all the potential trivial and basic questions and thank you for taking the time to answer them!
