Problem with Metal Roughness Parameter

Hello Radiance Community,

I'm having a problem with the metal material when it is both highly specular
and rough. In my room model, I have a vertical array of specular louvers at
the top of a glazed facade which bounce light onto the ceiling. The ceiling
for the first 6 m from the facade has a bumpy specular surface, using the
metal material. Material description:

void metal ceil_diff2
0
0
5 .856 .856 .856 .95 .125

The problem I'm seeing is that as I increase the roughness of the metal
ceiling from 0 (I've tried 0, .125, .2, and .9) the total amount of light in
the room increases significantly. This should not happen as the ceiling is
only redistributing the light that enters through the louver array. I found
that metal with 0 roughness produces essentially the same results as if I
used the mirror material on the ceiling, which we would expect. As I
increase the roughness, the area under the illuminance curve increases (to
absurd levels in the case of .9 roughness, but I know that .2 is about the
max for real objects). I tested the illuminance profile at the centerline
of the room as well as nearer to the side walls and saw the same pattern, so
I don't think it's a matter of undersampling the room. The only thing I
changed between cases was the roughness of the ceiling.

Can anyone think of what might be causing this effect? Are there any known
problems relating to the roughness of the metal material? I've searched the
archives, but didn't find an answer to my problem.

If you are interested in helping, here is a link to the graph of the
illuminance results for my test room. The values are the aggregate of
individual time steps from an annual simulation. The legend shows the
roughness in parentheses.

https://docs.google.com/leaf? <http://goog_618206457/>
id=0B69GQAj_Aw_QNzZhNWMwOTktZjRjOS00NWJiLThhMTUtODNlNDQxYmMyYjJi&hl=en&authkey=CIC91fUF<https://docs.google.com/leaf?id=0B69GQAj_Aw_QNzZhNWMwOTktZjRjOS00NWJiLThhMTUtODNlNDQxYmMyYjJi&hl=en&authkey=CIC91fUF>

Thanks very much for your help. I can send more information on my model if
requested.

-Kevin

···

--
Kevin Thuot
Research Assistant
Massachusetts Institute of Technology
[email protected]
(512) 497-8168

Hi Kevin!

I'm having a problem with the metal material when it is both highly
specular and rough. In my room model, I have a vertical array of
specular louvers at the top of a glazed facade which bounce light onto
the ceiling.

(...)

The problem I'm seeing is that as I increase the roughness of the metal
ceiling from 0 (I've tried 0, .125, .2, and .9) the total amount of
light in the room increases significantly.

I am just following your textual description as such:

1) direct sunlight gets specularly reflected from louvers to ceiling

2) light gets reflected from ceiling to sensor (and you have a very high specular reflectance of .95)

As long as the sun is not perfectly aligned perpendicular with your facade, you will probably also have some diffuse reflection over the wall, but that should not cause these effects.

Is it that surprising?

In case of zero roughness, all light reflected by the ceiling probably hits the back wall. Your sensor should probably only get hit by light reflected of the back wall (if there is any), light from the sky and maybe scattered light from the louvers.

In case of higher roughnesses, the distribution after the reflection on the ceiling gets wider - less light reaches the back wall, more light reaches the sensor. Seen from below, the ceiling would appear brighter (unless you are at the exact sun position where the sun would get reflected at by a mirror-like ceiling).

All this can be visualized by holding a mirror, a scattering mirror and a piece of paper in front of a (small) lamp in a dark room. The mirror will appear black - unless you find the exact one position where light gets reflected straight into your eyes. The scatter mirror will appear brighter when seen close to the mirror angle, paper will appear bright at any angle.

My guess that this is all the reason for your observation. Am I right?

Cheers, Lars.

If the models are already available it shouldn't be much work to create renderings of the different scenarios to see if you can confirm Lars suspicions.

···

In case of zero roughness, all light reflected by the ceiling probably
hits the back wall. Your sensor should probably only get hit by light
reflected of the back wall (if there is any), light from the sky and
maybe scattered light from the louvers.

In case of higher roughnesses, the distribution after the reflection on
the ceiling gets wider - less light reaches the back wall, more light
reaches the sensor. Seen from below, the ceiling would appear brighter
(unless you are at the exact sun position where the sun would get
reflected at by a mirror-like ceiling).

____________________________________________________________
Electronic mail messages entering and leaving Arup business
systems are scanned for acceptability of content and viruses