"Sensor Surfaces"

Dear all

in order to be able to "scan" a large surface which is located behind a
partially and non-uniformly transparent surface for "if" direct light hits,
I am looking for a way to have "Sensor Surfaces" in a Radiance model. I
would like to avoid having to define discrete sensor points as this would
either require an "infinite" number of such points or lead to too large gaps
in the information, IMO.

I have been unable to find any reference to such a possibility. If anybody
here has any ideas, I would greatly appreciate them.

Thanks and best regards

Achim

Yuanda-Europe

Hi Achim.

Radzilla has a 'falsecolor' material you could use.

With the standard Radiance you could use John Mardaljevic's 'stencil'
method to calculate rtrace origin points on the fly. The resolution of
your 'surface' then only depends on the resolution of your output
image. More details here:

http://www.radiance-online.org/pipermail/radiance-general/2007-March/004214.html

Regards,
Thomas

···

On Thu, Nov 19, 2009 at 1:47 PM, Achim Geissler <[email protected]> wrote:

Dear all

in order to be able to „scan“ a large surface which is located behind a
partially and non-uniformly transparent surface for “if” direct light hits,
I am looking for a way to have “Sensor Surfaces” in a Radiance model. I
would like to avoid having to define discrete sensor points as this would
either require an “infinite” number of such points or lead to too large gaps
in the information, IMO.

I have been unable to find any reference to such a possibility. If anybody
here has any ideas, I would greatly appreciate them.

Thanks and best regards

Achim

Yuanda-Europe

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

Hi Thomas

thanks for your quick reply. It took me a few days to try to understand the
possibilities you suggested.

The 'falsecolor' sounds quite interesting, but as far as I can tell it does
not allow "automatic" evaluation, i.e. create a file / files that can be
somehow scanned for "hit" or "no hit" per surface. My problem is, that
visually evaluating images after the simulation does not seem practical due
to form / size / repetitions.

And I'm afraid I didn't understand if the stencil method could really help,
but it seems evaluation (or result) is still a "picture" - again not a "hit"
or "no hit" information for a given surface. Actually, I didn't understand
the method, period (and haven't found the time to set up an example and play
around with it).

Something like
http://www.radiance-online.org/radiance-workshop7/Content/Augsburger/Germain
AugsburgerPresentation.pdf, page 8, but with "continuous" and not
"discrete" sensors that can be evaluated for hit-or-miss somehow is what I
have in mind.

I must admit of course (if not clear from the above, anyway) that my
knowledge of the possibilities of Radiance is rather limited.

Thanks
Achim

···

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Thomas
Bleicher
Sent: Donnerstag, 19. November 2009 15:08
To: Radiance general discussion
Subject: Re: [Radiance-general] "Sensor Surfaces"

Hi Achim.

Radzilla has a 'falsecolor' material you could use.

With the standard Radiance you could use John Mardaljevic's 'stencil'
method to calculate rtrace origin points on the fly. The resolution of
your 'surface' then only depends on the resolution of your output
image. More details here:

http://www.radiance-online.org/pipermail/radiance-general/2007-March/004214.
html

Regards,
Thomas

On Thu, Nov 19, 2009 at 1:47 PM, Achim Geissler <[email protected]> wrote:

Dear all

in order to be able to "scan" a large surface which is located behind a
partially and non-uniformly transparent surface for "if" direct light

hits,

I am looking for a way to have "Sensor Surfaces" in a Radiance model. I
would like to avoid having to define discrete sensor points as this would
either require an "infinite" number of such points or lead to too large

gaps

in the information, IMO.

I have been unable to find any reference to such a possibility. If anybody
here has any ideas, I would greatly appreciate them.

Thanks and best regards

Achim

Yuanda-Europe

_______________________________________________
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

Achim.

I'm afraid I don't quite understand what you want to do.

The 'falsecolor' sounds quite interesting, but as far as I can tell it does
not allow "automatic" evaluation, i.e. create a file / files that can be
somehow scanned for "hit" or "no hit" per surface. My problem is, that
visually evaluating images after the simulation does not seem practical due
to form / size / repetitions.

I recently implemented a script that checks for a hit and miss
of the sky at possible sun positions. I just used the rtrace output
option to produce the material name of the hit surface.

Example input:
0 0 0 0 0 1
0 0 0 0 1 1
0 0 0 1 1 1

Example output:
skyglow
grey_40
grey_40

If you use a unique material name you can afterwards parse the
output with a simple awk one-liner to get the number of hits. Is
that more what you need?

And I'm afraid I didn't understand if the stencil method could really help,
but it seems evaluation (or result) is still a "picture" - again not a "hit"
or "no hit" information for a given surface.

The output is a picture if you want rtrace to produce a picture.
It only depends on the options for rtrace. The nice thing about
the stencil method is that it generates ray origins and directions
over a surface, instead of a fixed size grid.

Something like
http://www.radiance-online.org/radiance-workshop7/Content/Augsburger/Germain
AugsburgerPresentation.pdf, page 8, but with "continuous" and not
"discrete" sensors that can be evaluated for hit-or-miss somehow is what I
have in mind.

Stencil could create the x,y,z,dx,dy,dz coordinates but only
for one half of the cylinder (the one facing you). And the
resolution would degrade towards the edges.

Perhaps you could outline your problem. Someone might have
another idea.

Regards,
Thomas

···

On Wed, Nov 25, 2009 at 7:40 AM, Achim Geissler <[email protected]> wrote:

Thomas

that script-solution sounds quite interesting and what you describe as what
you did with it sounds more like what I am looking for. The multitude of
possible switches for rtrace is quite a challenge, though.

What I want to do is this: I have a large building with external shading
that has a non-trivial geometric form. This shading is to be opened (to a
certain degree) at a given azimuth only if no direct radiation which would
hit the facade can get past in this location. So the task is to let the sun
travel around the building and for each position of interest check if any
direct hits can be found on any surface of the building façade behind the
shading.

Your suggestion seems to imply that this should be possible with rtrace.
Does setting -dr 0 mean that only direct beams are considered by rtrace? Or
does one need a good combination of -ds, -dt and -dc? The sun is the only
"active" source of light. I would set reflection of all shading surfaces to
Zero.

I am not sure what your input example describes. Having the material names
of "hit" surfaces given would seem O.K. if this can be constrained to
"direct" only, as the goal is to have "no direct hit".

Regards
Achim

···

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Thomas
Bleicher
Sent: Mittwoch, 25. November 2009 18:42
To: Radiance general discussion
Subject: Re: [Radiance-general] "Sensor Surfaces"

Achim.

I'm afraid I don't quite understand what you want to do.

On Wed, Nov 25, 2009 at 7:40 AM, Achim Geissler <[email protected]> wrote:

The 'falsecolor' sounds quite interesting, but as far as I can tell it

does

not allow "automatic" evaluation, i.e. create a file / files that can be
somehow scanned for "hit" or "no hit" per surface. My problem is, that
visually evaluating images after the simulation does not seem practical

due

to form / size / repetitions.

I recently implemented a script that checks for a hit and miss
of the sky at possible sun positions. I just used the rtrace output
option to produce the material name of the hit surface.

Example input:
0 0 0 0 0 1
0 0 0 0 1 1
0 0 0 1 1 1

Example output:
skyglow
grey_40
grey_40

If you use a unique material name you can afterwards parse the
output with a simple awk one-liner to get the number of hits. Is
that more what you need?

And I'm afraid I didn't understand if the stencil method could really

help,

but it seems evaluation (or result) is still a "picture" - again not a

"hit"

or "no hit" information for a given surface.

The output is a picture if you want rtrace to produce a picture.
It only depends on the options for rtrace. The nice thing about
the stencil method is that it generates ray origins and directions
over a surface, instead of a fixed size grid.

Something like

http://www.radiance-online.org/radiance-workshop7/Content/Augsburger/Germain

AugsburgerPresentation.pdf, page 8, but with "continuous" and not
"discrete" sensors that can be evaluated for hit-or-miss somehow is what I
have in mind.

Stencil could create the x,y,z,dx,dy,dz coordinates but only
for one half of the cylinder (the one facing you). And the
resolution would degrade towards the edges.

Perhaps you could outline your problem. Someone might have
another idea.

Regards,
Thomas

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

Hi Achim.

The description of your project makes the situation clearer.

What I want to do is this: I have a large building with external shading
that has a non-trivial geometric form. This shading is to be opened (to a
certain degree) at a given azimuth only if no direct radiation which would
hit the facade can get past in this location. So the task is to let the sun
travel around the building and for each position of interest check if any
direct hits can be found on any surface of the building façade behind the
shading.

Both options I mentioned could be used here:

A - stencil:
1) based on a view of the building facade you generate x,y,z coordinates
    (sensor points). The resolution of your sensor points depends on the
    view and image specification; it has it's limits but compared to a grid
    it should be better. You can probably restrict the area to test to a
    representive patch.
2) the direction is the surface normal at that point
3) with -ab 0 and only the sun (no sky dome) you create an image
4) If any of your image points is bright (not 0) you have a hit. You can
    use pvalue to do this automatically or just look at the image.

B - search for material:
1) define grid resolution to apply to building facade
2) use sun direction as test direction
3) use rtrace with -os (surface) or -oM (material) and without -I+
4) if you get a "sun" or "solar" for any of the points you have a hit

C - combined method:
1) from stencil to create the grid points
2), 3), 4) as B

Your suggestion seems to imply that this should be possible with rtrace.
Does setting -dr 0 mean that only direct beams are considered by rtrace? Or
does one need a good combination of -ds, -dt and -dc? The sun is the only
"active" source of light. I would set reflection of all shading surfaces to
Zero.

In the second case all these settings do not matter. We are using the
output format spec to let rtrace report about the material it hits instead
of the radiance of that surface. For the stencil method I do not think
these settings are an issue because you only have the sun in the sky
and a single light source is tested automatically by rtrace. I might be
wrong, though.

I am not sure what your input example describes. Having the material names
of "hit" surfaces given would seem O.K. if this can be constrained to
"direct" only, as the goal is to have "no direct hit".

In the example I "tested" 3 directions from point (0,0,0).
In an rtrace output you would have 3 lines of materials.
Assuming that the sky material is called "skyglow" only
the first ray actually hits the sky. You can name the materials
as you like, of course. In fact, I don't use a sky at all and
look for "*" in the output which rtrace reports if it doesn't hit
anything.

Regards,
Thomas

···

On Thu, Nov 26, 2009 at 9:52 AM, Achim Geissler <[email protected]> wrote:

Hi Thomas

thank you very much! I am beginning to see the light, so to speak ;o)

Regards
Achim

···

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Thomas
Bleicher
Sent: Donnerstag, 26. November 2009 13:58
To: Radiance general discussion
Subject: Re: [Radiance-general] "Sensor Surfaces"

Hi Achim.

The description of your project makes the situation clearer.

On Thu, Nov 26, 2009 at 9:52 AM, Achim Geissler <[email protected]> wrote:

What I want to do is this: I have a large building with external shading
that has a non-trivial geometric form. This shading is to be opened (to a
certain degree) at a given azimuth only if no direct radiation which would
hit the facade can get past in this location. So the task is to let the

sun

travel around the building and for each position of interest check if any
direct hits can be found on any surface of the building façade behind the
shading.

Both options I mentioned could be used here:

A - stencil:
1) based on a view of the building facade you generate x,y,z coordinates
    (sensor points). The resolution of your sensor points depends on the
    view and image specification; it has it's limits but compared to a grid
    it should be better. You can probably restrict the area to test to a
    representive patch.
2) the direction is the surface normal at that point
3) with -ab 0 and only the sun (no sky dome) you create an image
4) If any of your image points is bright (not 0) you have a hit. You can
    use pvalue to do this automatically or just look at the image.

B - search for material:
1) define grid resolution to apply to building facade
2) use sun direction as test direction
3) use rtrace with -os (surface) or -oM (material) and without -I+
4) if you get a "sun" or "solar" for any of the points you have a hit

C - combined method:
1) from stencil to create the grid points
2), 3), 4) as B

Your suggestion seems to imply that this should be possible with rtrace.
Does setting -dr 0 mean that only direct beams are considered by rtrace?

Or

does one need a good combination of -ds, -dt and -dc? The sun is the only
"active" source of light. I would set reflection of all shading surfaces

to

Zero.

In the second case all these settings do not matter. We are using the
output format spec to let rtrace report about the material it hits instead
of the radiance of that surface. For the stencil method I do not think
these settings are an issue because you only have the sun in the sky
and a single light source is tested automatically by rtrace. I might be
wrong, though.

I am not sure what your input example describes. Having the material names
of "hit" surfaces given would seem O.K. if this can be constrained to
"direct" only, as the goal is to have "no direct hit".

In the example I "tested" 3 directions from point (0,0,0).
In an rtrace output you would have 3 lines of materials.
Assuming that the sky material is called "skyglow" only
the first ray actually hits the sky. You can name the materials
as you like, of course. In fact, I don't use a sky at all and
look for "*" in the output which rtrace reports if it doesn't hit
anything.

Regards,
Thomas

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