I am doing some analyses on reflectance values and I have just started using the 5-phase method for those.
At first I thought the reflectances assigned to the model would not matter for the part that is specific to the 5-phase, as the model is assigned a black material, but comparing the resulting matrices it looks like the Sun Coefficient Matrix (.dsc) changes with the original materials’ reflectances (and it’s the only matrix that change). And that is exactly the matrix that is taking most of the computational effort (about 4 hours I think - the same model with the 2-phase method takes 5 minutes, glazing is all simple clear glass).
Am I doing something wrong or is this the normal behaviour? The data I get in the end are plausible. Could you explain why the Sun Coefficient Matrix needs to take into account the model reflectances at all?
I hope I managed to explain the issue somehow... Thanks for your help!
From: giulioandrea antonutto foi <geotrupes@me.com>
Subject: Re: [Radiance-general] 5-phase method model reflectance
Date: 17 June 2015 21:53:57 BST
To: Radiance general discussion <radiance-general@radiance-online.org>
Off topic completely…
your setting for dj and st are a bit odd. maybe interesting to know if there is a reason.
A jitter of 100% will lead to aiming failure?
and also setting a specular threshold to 100% will mean no specular sampling?
unless there is a reason? new radiance settings meaning different things?
just a curiosity.
G.
At first I thought the reflectances assigned to the model would not matter for the part that is specific to the 5-phase, as the model is assigned a black material, but comparing the resulting matrices it looks like the Sun Coefficient Matrix (.dsc) changes with the original materials’ reflectances (and it’s the only matrix that change). And that is exactly the matrix that is taking most of the computational effort (about 4 hours I think - the same model with the 2-phase method takes 5 minutes, glazing is all simple clear glass).
Off topic completely…
your setting for dj and st are a bit odd. maybe interesting to know if there is a reason.
A jitter of 100% will lead to aiming failure?
and also setting a specular threshold to 100% will mean no specular sampling?
unless there is a reason? new radiance settings meaning different things?
just a curiosity.
G.
At first I thought the reflectances assigned to the model would not matter for the part that is specific to the 5-phase, as the model is assigned a black material, but comparing the resulting matrices it looks like the Sun Coefficient Matrix (.dsc) changes with the original materials’ reflectances (and it’s the only matrix that change). And that is exactly the matrix that is taking most of the computational effort (about 4 hours I think - the same model with the 2-phase method takes 5 minutes, glazing is all simple clear glass).
The direct sun coefficient matrix only includes materials in your window
system, assuming you're using pass-through BSDF in you model. So it
shouldn't change when you change other materials. Maybe check that your
materials are actually black using rvu?
There are ways to reduce the amount of time the direct sun coefficient
matrix takes, the most useful is to delete the suns in suns.rad that will
never be used. So if you don't plan on changing the orientation or latitude
in your simulations you can delete suns that fall outside of your solar
band. Also if you're only doing hourly analysis, you could delete suns
between hourly analemas.
I am doing some analyses on reflectance values and I have just started
using the 5-phase method for those.
At first I thought the reflectances assigned to the model would not
matter for the part that is specific to the 5-phase, as the model is
assigned a black material, but comparing the resulting matrices it looks
like the Sun Coefficient Matrix (.dsc) changes with the original materials’
reflectances (and it’s the only matrix that change). And that is exactly
the matrix that is taking most of the computational effort (about 4 hours I
think - the same model with the 2-phase method takes 5 minutes, glazing is
all simple clear glass).
Am I doing something wrong or is this the normal behaviour? The data I
get in the end are plausible. Could you explain why the Sun Coefficient
Matrix needs to take into account the model reflectances at all?
I hope I managed to explain the issue somehow... Thanks for your help!
Greg (and David G-M) changed the way direct jittering works for round
sources so that there are no longer aiming failures (
).
and also setting a specular threshold to 100% will mean no specular
sampling?
You want to turn specular sampling off for the direct sun coefficient
matrix to avoid double counting in the 5-phase method.
Best,
Andy
···
On Wed, Jun 17, 2015 at 1:53 PM, giulioandrea antonutto foi < geotrupes@me.com> wrote:
Off topic completely…
your setting for dj and st are a bit odd. maybe interesting to know if
there is a reason.
A jitter of 100% will lead to aiming failure?
and also setting a specular threshold to 100% will mean no specular
sampling?
unless there is a reason? new radiance settings meaning different things?
just a curiosity.
G.
At first I thought the reflectances assigned to the model would not matter
for the part that is specific to the 5-phase, as the model is assigned a
black material, but comparing the resulting matrices it looks like the Sun
Coefficient Matrix (.dsc) changes with the original materials’ reflectances
(and it’s the only matrix that change). And that is exactly the matrix that
is taking most of the computational effort (about 4 hours I think - the
same model with the 2-phase method takes 5 minutes, glazing is all simple
clear glass).
and also setting a specular threshold to 100% will mean no specular sampling?
You want to turn specular sampling off for the direct sun coefficient matrix to avoid double counting in the 5-phase method.
Best,
Andy
On Wed, Jun 17, 2015 at 1:53 PM, giulioandrea antonutto foi <geotrupes@me.com> wrote:
Off topic completely…
your setting for dj and st are a bit odd. maybe interesting to know if there is a reason.
A jitter of 100% will lead to aiming failure?
and also setting a specular threshold to 100% will mean no specular sampling?
unless there is a reason? new radiance settings meaning different things?
just a curiosity.
G.
At first I thought the reflectances assigned to the model would not matter for the part that is specific to the 5-phase, as the model is assigned a black material, but comparing the resulting matrices it looks like the Sun Coefficient Matrix (.dsc) changes with the original materials’ reflectances (and it’s the only matrix that change). And that is exactly the matrix that is taking most of the computational effort (about 4 hours I think - the same model with the 2-phase method takes 5 minutes, glazing is all simple clear glass).
and also setting a specular threshold to 100% will mean no specular sampling?
You want to turn specular sampling off for the direct sun coefficient matrix to avoid double counting in the 5-phase method.
Best,
Andy
On Wed, Jun 17, 2015 at 1:53 PM, giulioandrea antonutto foi <geotrupes@me.com> wrote:
Off topic completely…
your setting for dj and st are a bit odd. maybe interesting to know if there is a reason.
A jitter of 100% will lead to aiming failure?
and also setting a specular threshold to 100% will mean no specular sampling?
unless there is a reason? new radiance settings meaning different things?
just a curiosity.
G.
At first I thought the reflectances assigned to the model would not matter for the part that is specific to the 5-phase, as the model is assigned a black material, but comparing the resulting matrices it looks like the Sun Coefficient Matrix (.dsc) changes with the original materials’ reflectances (and it’s the only matrix that change). And that is exactly the matrix that is taking most of the computational effort (about 4 hours I think - the same model with the 2-phase method takes 5 minutes, glazing is all simple clear glass).
and also setting a specular threshold to 100% will mean no specular
sampling?
You want to turn specular sampling off for the direct sun coefficient
matrix to avoid double counting in the 5-phase method.
Best,
Andy
On Wed, Jun 17, 2015 at 1:53 PM, giulioandrea antonutto foi < > geotrupes@me.com> wrote:
Off topic completely…
your setting for dj and st are a bit odd. maybe interesting to know if
there is a reason.
A jitter of 100% will lead to aiming failure?
and also setting a specular threshold to 100% will mean no specular
sampling?
unless there is a reason? new radiance settings meaning different things?
just a curiosity.
G.
At first I thought the reflectances assigned to the model would not
matter for the part that is specific to the 5-phase, as the model is
assigned a black material, but comparing the resulting matrices it looks
like the Sun Coefficient Matrix (.dsc) changes with the original materials’
reflectances (and it’s the only matrix that change). And that is exactly
the matrix that is taking most of the computational effort (about 4 hours I
think - the same model with the 2-phase method takes 5 minutes, glazing is
all simple clear glass).
ok, if you all insist I will not quit
but beware, I may ask the odd thing here and there…
thanks
G
···
On 18 Jun 2015, at 02:14, Andrew McNeil <amcneil@lbl.gov> wrote:
Noooooooooo!
Don't quit.
On Wed, Jun 17, 2015 at 5:29 PM, Greg Ward <gregoryjward@gmail.com> wrote:
Don't say that, Giulio -- I would have to quit, too! I was also puzzled by the -st 1 setting.... (I still need to learn Andy's 5-phase method.)
-Greg
From: G <antonutto@yahoo.it>
Subject: Re: [Radiance-general] 5-phase method model reflectance
Date: June 17, 2015 5:15:56 PM PDT
Thanks Andy!
There is a time to quit boxing…
Maybe it is my time today…
G
On 17 Jun 2015, at 22:25, Andrew McNeil <amcneil@lbl.gov> wrote:
and also setting a specular threshold to 100% will mean no specular sampling?
You want to turn specular sampling off for the direct sun coefficient matrix to avoid double counting in the 5-phase method.
Best,
Andy
On Wed, Jun 17, 2015 at 1:53 PM, giulioandrea antonutto foi <geotrupes@me.com> wrote:
Off topic completely…
your setting for dj and st are a bit odd. maybe interesting to know if there is a reason.
A jitter of 100% will lead to aiming failure?
and also setting a specular threshold to 100% will mean no specular sampling?
unless there is a reason? new radiance settings meaning different things?
just a curiosity.
G.
At first I thought the reflectances assigned to the model would not matter for the part that is specific to the 5-phase, as the model is assigned a black material, but comparing the resulting matrices it looks like the Sun Coefficient Matrix (.dsc) changes with the original materials’ reflectances (and it’s the only matrix that change). And that is exactly the matrix that is taking most of the computational effort (about 4 hours I think - the same model with the 2-phase method takes 5 minutes, glazing is all simple clear glass).
