Strange halo artifact in rendering

Folks,

I am stumped. There is a halo around a black box on a grey background
in a rendering, and I can't figure out how to get rid of it (because I
can't imagine that it's real).

I am preparing renderings for an art submission, and I am simulating a
little model mountain inside a small flat black frame on a light grey
wall, in a room with an array of ceiling lights and one spot. Here is
what happens (rpict and mountain-removed rvu):


The black material is plastic with 0.0 specularity.
The halo is independent of the ambient parameters.
The halo is independent of -lw refinement.
The halo is still there if the black box touches the wall or is offset a little.
The halo doesn't go away when I render at high resolution.

Can anyone tell me what is going on here? Unzip and run "make" on:
http://markjstock.org/transfer/whatGives.zip

Mark

Hi Mark,

I tried to run your example but I failed. Some files seem to be missing.
What about increasing the roughness of your black frame?

Best Kai

  Freundliche Grüße
  Kai Babetzki
  Dipl.-Phys. Kai Babetzki
  Mailto:[email protected] phone: +49-711-67976-288
  http://www.transsolar.com fax: +49-711-67976-11

···

---------------------------------------------------------------------------------------------------------------------------------------
  Transsolar KlimaEngineering
  Technologien für energieeffizientes Bauen und Nutzerkomfort in Gebäuden
  Strategies for energy efficient design and thermal comfort in buildings
  Stuttgart - München - New York
  Transsolar Energietechnik GmbH
  Curiestrasse 2- 70563 Stuttgart
  Geschäftsführer :
  Diplom-Ingenieur Matthias Schuler, Diplom-Ingenieur Thomas Auer, Diplom-Physiker Stefan Holst
  Eingetragen beim Amtsgericht Stuttgart HRB 23347 USt-IdNr.: 99073/00911
  ---------------------------------------------------------------------------------------------------------------------------------------

-----Ursprüngliche Nachricht-----
Von: Mark Stock [mailto:[email protected]]
Gesendet: Freitag, 15. März 2013 05:56
An: [email protected]
Betreff: [Radiance-general] Strange halo artifact in rendering

Folks,

I am stumped. There is a halo around a black box on a grey background in a rendering, and I can't figure out how to get rid of it (because I can't imagine that it's real).

I am preparing renderings for an art submission, and I am simulating a little model mountain inside a small flat black frame on a light grey wall, in a room with an array of ceiling lights and one spot. Here is what happens (rpict and mountain-removed rvu):


The black material is plastic with 0.0 specularity.
The halo is independent of the ambient parameters.
The halo is independent of -lw refinement.
The halo is still there if the black box touches the wall or is offset a little.
The halo doesn't go away when I render at high resolution.

Can anyone tell me what is going on here? Unzip and run "make" on:
http://markjstock.org/transfer/whatGives.zip

Mark

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

Kai,

Thanks for checking. Which files were missing?

Changed black frame to "0.03 0.03 0.03 0.0 0.2" and no change.

Mark

···

On 3/15/13, Kai Babetzki <[email protected]> wrote:

Hi Mark,

I tried to run your example but I failed. Some files seem to be missing.
What about increasing the roughness of your black frame?

Best Kai

  Freundliche Grüße
  Kai Babetzki
  Dipl.-Phys. Kai Babetzki
  Mailto:[email protected] phone: +49-711-67976-288
  http://www.transsolar.com fax: +49-711-67976-11
  ---------------------------------------------------------------------------------------------------------------------------------------
  Transsolar KlimaEngineering
  Technologien für energieeffizientes Bauen und Nutzerkomfort in Gebäuden
  Strategies for energy efficient design and thermal comfort in buildings
  Stuttgart - München - New York
  Transsolar Energietechnik GmbH
  Curiestrasse 2- 70563 Stuttgart
  Geschäftsführer :
  Diplom-Ingenieur Matthias Schuler, Diplom-Ingenieur Thomas Auer,
Diplom-Physiker Stefan Holst
  Eingetragen beim Amtsgericht Stuttgart HRB 23347 USt-IdNr.: 99073/00911
  ---------------------------------------------------------------------------------------------------------------------------------------

-----Ursprüngliche Nachricht-----
Von: Mark Stock [mailto:[email protected]]
Gesendet: Freitag, 15. März 2013 05:56
An: [email protected]
Betreff: [Radiance-general] Strange halo artifact in rendering

Folks,

I am stumped. There is a halo around a black box on a grey background in a
rendering, and I can't figure out how to get rid of it (because I can't
imagine that it's real).

I am preparing renderings for an art submission, and I am simulating a
little model mountain inside a small flat black frame on a light grey wall,
in a room with an array of ceiling lights and one spot. Here is what happens
(rpict and mountain-removed rvu):

http://markjstock.org/transfer/whyTheHaloMaster.png
http://markjstock.org/transfer/whyTheHalo.png

The black material is plastic with 0.0 specularity.
The halo is independent of the ambient parameters.
The halo is independent of -lw refinement.
The halo is still there if the black box touches the wall or is offset a
little.
The halo doesn't go away when I render at high resolution.

Can anyone tell me what is going on here? Unzip and run "make" on:
http://markjstock.org/transfer/whatGives.zip

Mark

_______________________________________________
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

Hey Mark,

What are your parameters? Have you taken a look at -dr, -dc, -dt?

-Jack

···

--
Jack de Valpine
President

Visarc Incorporated
www.visarc.com

channeling technology for superior design and construction

On 3/15/2013 8:46 AM, Mark Stock wrote:

Kai,

Thanks for checking. Which files were missing?

Changed black frame to "0.03 0.03 0.03 0.0 0.2" and no change.

Mark

On 3/15/13, Kai Babetzki <[email protected]> wrote:

Hi Mark,

I tried to run your example but I failed. Some files seem to be missing.
What about increasing the roughness of your black frame?

Best Kai

  Freundliche Grüße
  Kai Babetzki
  Dipl.-Phys. Kai Babetzki
  Mailto:[email protected] phone: +49-711-67976-288
  http://www.transsolar.com fax: +49-711-67976-11
  ---------------------------------------------------------------------------------------------------------------------------------------
  Transsolar KlimaEngineering
  Technologien für energieeffizientes Bauen und Nutzerkomfort in Gebäuden
  Strategies for energy efficient design and thermal comfort in buildings
  Stuttgart - München - New York
  Transsolar Energietechnik GmbH
  Curiestrasse 2- 70563 Stuttgart
  Geschäftsführer :
  Diplom-Ingenieur Matthias Schuler, Diplom-Ingenieur Thomas Auer,
Diplom-Physiker Stefan Holst
  Eingetragen beim Amtsgericht Stuttgart HRB 23347 USt-IdNr.: 99073/00911
  ---------------------------------------------------------------------------------------------------------------------------------------

-----Ursprüngliche Nachricht-----
Von: Mark Stock [mailto:[email protected]]
Gesendet: Freitag, 15. März 2013 05:56
An: [email protected]
Betreff: [Radiance-general] Strange halo artifact in rendering

Folks,

I am stumped. There is a halo around a black box on a grey background in a
rendering, and I can't figure out how to get rid of it (because I can't
imagine that it's real).

I am preparing renderings for an art submission, and I am simulating a
little model mountain inside a small flat black frame on a light grey wall,
in a room with an array of ceiling lights and one spot. Here is what happens
(rpict and mountain-removed rvu):

http://markjstock.org/transfer/whyTheHaloMaster.png
http://markjstock.org/transfer/whyTheHalo.png

The black material is plastic with 0.0 specularity.
The halo is independent of the ambient parameters.
The halo is independent of -lw refinement.
The halo is still there if the black box touches the wall or is offset a
little.
The halo doesn't go away when I render at high resolution.

Can anyone tell me what is going on here? Unzip and run "make" on:
http://markjstock.org/transfer/whatGives.zip

Mark

_______________________________________________
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

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

Sorry I did not mean -dr (thinking of a different thread), so -ds, -dc, -dt...

···

--
Jack de Valpine
President

Visarc Incorporated
www.visarc.com

channeling technology for superior design and construction

On 3/15/2013 8:48 AM, Jack de Valpine wrote:

Hey Mark,

What are your parameters? Have you taken a look at -dr, -dc, -dt?

-Jack

Jack,

Sorry. Reducing -dc -dt and -ds doesn't seem to do anything.

I learned this morning that the halo is coming from the array of ceiling lights and not the spotlight.

I also discovered that the halo gets smaller as the black box is moved away from the wall.

And the antimatter cutout that turns the box into a frame doesn't affect the halo.

Mark

···

On Fri, 15 Mar 2013, Jack de Valpine wrote:

Sorry I did not mean -dr (thinking of a different thread), so -ds, -dc, -dt...

--
Jack de Valpine
President

Visarc Incorporated
www.visarc.com

channeling technology for superior design and construction

On 3/15/2013 8:48 AM, Jack de Valpine wrote:

Hey Mark,

What are your parameters? Have you taken a look at -dr, -dc, -dt?

-Jack

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

Hi Mark,

Yeah, poking around a bit I see the direct parameter don't seem to have much impact. On the other hand I found that if I set -ar to 256 then the halo disappears.

-Jack

···

--
Jack de Valpine
President

Visarc Incorporated
www.visarc.com

channeling technology for superior design and construction

On 3/15/2013 9:57 AM, Mark Stock wrote:

Jack,

Sorry. Reducing -dc -dt and -ds doesn't seem to do anything.

I learned this morning that the halo is coming from the array of ceiling lights and not the spotlight.

I also discovered that the halo gets smaller as the black box is moved away from the wall.

And the antimatter cutout that turns the box into a frame doesn't affect the halo.

Mark

On Fri, 15 Mar 2013, Jack de Valpine wrote:

Sorry I did not mean -dr (thinking of a different thread), so -ds, -dc, -dt...

--
Jack de Valpine
President

Visarc Incorporated
www.visarc.com

channeling technology for superior design and construction

On 3/15/2013 8:48 AM, Jack de Valpine wrote:

Hey Mark,

What are your parameters? Have you taken a look at -dr, -dc, -dt?

-Jack

_______________________________________________
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

Jack,

Excellent! Thank you. I am surprised that -ar has any effect when sampling is turned off (-aa 0). I didn't even try to modify -ar because I didn't think it would do anything.

Mark

···

On Fri, 15 Mar 2013, Jack de Valpine wrote:

Hi Mark,

Yeah, poking around a bit I see the direct parameter don't seem to have much impact. On the other hand I found that if I set -ar to 256 then the halo disappears.

-Jack

--
Jack de Valpine
President

Visarc Incorporated
www.visarc.com

channeling technology for superior design and construction

On 3/15/2013 9:57 AM, Mark Stock wrote:

Jack,

Sorry. Reducing -dc -dt and -ds doesn't seem to do anything.

I learned this morning that the halo is coming from the array of ceiling lights and not the spotlight.

I also discovered that the halo gets smaller as the black box is moved away from the wall.

And the antimatter cutout that turns the box into a frame doesn't affect the halo.

Mark

On Fri, 15 Mar 2013, Jack de Valpine wrote:

Sorry I did not mean -dr (thinking of a different thread), so -ds, -dc, -dt...

--
Jack de Valpine
President

Visarc Incorporated
www.visarc.com

channeling technology for superior design and construction

On 3/15/2013 8:48 AM, Jack de Valpine wrote:

Hey Mark,

What are your parameters? Have you taken a look at -dr, -dc, -dt?

-Jack

_______________________________________________
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

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

I learned this morning that the halo is coming from the array of ceiling
lights and not the spotlight.

That was pretty clear to me, in looking at the rendering. The multiple
edges evident particularly from the bottom of the frame are the giveaway.
You are using Radiance light objects, and these always give these
exceedingly (unbelievably) sharp edges to the pools of light that they
cast. Radiance lights based on a photometry file tend to have softer
edges, even those based on directional lamps.

Jack's trick worked because even tho you have -aa at zero, you have -ab 2
which will still cause the ambient calculation to be run; the increased
-ar has the effect of softening the halo, I think. -aa 0 simply turns off
interpolation, not the ambient calculation altogether.

I think.

- Rob

That doesn't sit well with me. I do want the ambient calculation, but "-aa 0" is supposed to turn caching off---or run a fresh new hemisphere sampling for every image pixel (assuming -ps 1). So why would it interpolate from some brighter ambient point in the dark area *under the frame* instead of probing outward for the incident light? Ambient resolution should have no part of that.

M

···

On Fri, 15 Mar 2013, Guglielmetti, Robert wrote:

I learned this morning that the halo is coming from the array of ceiling
lights and not the spotlight.

That was pretty clear to me, in looking at the rendering. The multiple
edges evident particularly from the bottom of the frame are the giveaway.
You are using Radiance light objects, and these always give these
exceedingly (unbelievably) sharp edges to the pools of light that they
cast. Radiance lights based on a photometry file tend to have softer
edges, even those based on directional lamps.

Jack's trick worked because even tho you have -aa at zero, you have -ab 2
which will still cause the ambient calculation to be run; the increased
-ar has the effect of softening the halo, I think. -aa 0 simply turns off
interpolation, not the ambient calculation altogether.

I think.

- Rob

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

unless you did already,
try:
-as 0
and -ad as you please
G

···

On 15 Mar 2013, at 04:56, Mark Stock <[email protected]> wrote:

Folks,

I am stumped. There is a halo around a black box on a grey background
in a rendering, and I can't figure out how to get rid of it (because I
can't imagine that it's real).

I am preparing renderings for an art submission, and I am simulating a
little model mountain inside a small flat black frame on a light grey
wall, in a room with an array of ceiling lights and one spot. Here is
what happens (rpict and mountain-removed rvu):

http://markjstock.org/transfer/whyTheHaloMaster.png
http://markjstock.org/transfer/whyTheHalo.png

The black material is plastic with 0.0 specularity.
The halo is independent of the ambient parameters.
The halo is independent of -lw refinement.
The halo is still there if the black box touches the wall or is offset a little.
The halo doesn't go away when I render at high resolution.

Can anyone tell me what is going on here? Unzip and run "make" on:
http://markjstock.org/transfer/whatGives.zip

Mark

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