# Glare and evaluation of lightshelves

Hi there!

I'm using DR 2.0 and I've been making parametric
calculations comparing daylight systems -a room with high
reflectances, and the same room but adding in one case a
lightshelf and in another an anidolic reflector. The room
faces south, and I am evaluating them through daylight
glare index (dgi).

To give more details, the room measures 8x12 meters, with a
ceiling height of 2.70 meters. The window is 8x1.70m
(placed on the 8m side).

The materials: ceiling: brushed aluminium (r=80%), walls:
beige (r=65%) and floor: blue (r=20%). These reflectances
were taken from recommendations from literature.

The light shelf has a length of 1m and is also made from
brushed aluminium and is placed 2.1 meters from the floor.
(half is inside the room and half outside).

I generated pictures with winrview, both looking into the
window and looking sideways, and then used the two main
glare routines available (findglare and glarendx).

According to other studies (including those of people who
have used Radiance) the lightshelf should have a lower
glare index than the room without any addition, but to my
surprise and in the two picture cases, it is the opposite!
(absolute illuminance levels are OK, though)

In the beggining I thought that my ceiling was too specular
(50%) but then I changed it to another with similar
reflectance but no specularity... and the results are still
the same!

What could be wrong?

Carlos

···

_________________________________________________________
Do You Yahoo!?
Informaci�n de Estados Unidos y Am�rica Latina, en Yahoo! Noticias.
Vis�tanos en http://noticias.espanol.yahoo.com

Hi Carlos,

Glare calculations are quite tricky, because they are sensitive to the exact view you choose to calculate them, and there are a number of choices for glare indices as well, none of them terribly well-suited to daylight. (Not in glarendx, at least.)

It is impossible to determine anything from your description of your situation. Are you giving fisheye views to findglare? What do these images look like? Are you giving findglare an octree so that it may call rtrace?

And for the group, has anyone determined that findglare works correctly in Desktop Radiance?

-Greg

···

From: Carlos Ochoa <[email protected]>
Date: Sun Nov 16, 2003 5:09:15 AM US/Pacific

Hi there!

I'm using DR 2.0 and I've been making parametric
calculations comparing daylight systems -a room with high
reflectances, and the same room but adding in one case a
lightshelf and in another an anidolic reflector. The room
faces south, and I am evaluating them through daylight
glare index (dgi).

To give more details, the room measures 8x12 meters, with a
ceiling height of 2.70 meters. The window is 8x1.70m
(placed on the 8m side).

The materials: ceiling: brushed aluminium (r=80%), walls:
beige (r=65%) and floor: blue (r=20%). These reflectances
were taken from recommendations from literature.

The light shelf has a length of 1m and is also made from
brushed aluminium and is placed 2.1 meters from the floor.
(half is inside the room and half outside).

I generated pictures with winrview, both looking into the
window and looking sideways, and then used the two main
glare routines available (findglare and glarendx).

According to other studies (including those of people who
have used Radiance) the lightshelf should have a lower
glare index than the room without any addition, but to my
surprise and in the two picture cases, it is the opposite!
(absolute illuminance levels are OK, though)

In the beggining I thought that my ceiling was too specular
(50%) but then I changed it to another with similar
reflectance but no specularity... and the results are still
the same!

What could be wrong?

Carlos

Hi Greg,

The pictures were generated using the fish-eye view, they
are looking from the middle of one of the long walls of the
room to the other (the expected position of the user). So I
have the window on one side of the picture and the bottom
of the room on the other.

I give the findglare both the octree and the picture name,
but I get the same results just as if specifying only the
picture. By the way, just giving findglare the octree and
direction of the view gives an error message in DR 2.0. I
have also increased the number of ambient bounces, but it
doesn't change, either.

Carlos

···

_________________________________________________________
Do You Yahoo!?
Informaci�n de Estados Unidos y Am�rica Latina, en Yahoo! Noticias.
Vis�tanos en http://noticias.espanol.yahoo.com

Thanks, Phillip. I have corrected this bug in the calculation of dgr in glarendx.c. I was not aware of it. It is fixed for the next release.

-Greg

···

From: "Phillip Greenup" <[email protected]>
Date: Tue Nov 18, 2003 2:08:18 PM US/Pacific

Hi Carlos,

Are you predicting glare using the DGI? if you are, there may be some errors produced by potential glare sources behind the field of view.

findglare identifies potential glare sources. then glarendx -t dgi takes these sources and calculates glare indices in a range of directions. take a look at the predicted dgi's at adjacent directions. if there is an obvious jump in dgi between adjacent directions, then you probably have a glare source behind the field of view. To fix, simply remove the offending glare source from the file output by findglare, and re-run glarendx -t dgi. if that doesn't make any sense, let me know and i'll send you a spreadsheet i developed to identify problematic glare sources.

For Greg, or whoever else might be interested, the problem is in the calculation of dgi's omega in glarendx.c. where the glare source is behind the field of view, posindex returns -1, p becomes 0 and omega is much bigger than adjacent levels. this means the glare source is artificially inflated, and the glare index jumps up.

Phil.

Hi Carlos,

Unfortunately, I don't have a lot of time to play around with this right now. It looks on first glance that your ceiling is highly specular -- is it made of metal? I can't understand these sorts of reflections off any standard painted surface.

As for the bug in glarendx, I have fixed it for tomorrow's HEAD, but I don't think this is your problem as you only use a 180 degree field of view in your fisheye image, leaving findglare unable to fully characterize the window. I would think it better if you increased your field of view. Also, I'm not sure that the various glare index calculations are stable for sources that are at right angles to the view direction. I recommend that you use the -ga option to findglare to generate vertical illuminances at several angles, then use these in glarendx to get a function of glare with respect to view direction. This way, you'll be able to see how sensitive the calculation is to viewing direction. It will only work with a wider view, however, so you will need to use the view type -vta rather than -vth.

-Greg

···

From: Carlos Ochoa <[email protected]>
Date: Wed Nov 19, 2003 4:47:15 AM US/Pacific

Hi Greg,

This is Carlos again, I'm attaching with these email the
picture files and the octrees of the cases I was describing
for the radiance online group, since I think it is a bit
difficult to describe them exactly with words.

I read also the other comment about the bug in DGI arising
from the source being behind the field of view... could it
be?

Many thanks,

Carlos Ochoa

Hi Phillip,

Thanks for the email about glare, I tried to identify the glare source as you said, but I think if you send the spreadsheet you mentioned could be very useful!

Many thanks,
Carlos

···

Phillip Greenup <[email protected]> wrote:
Hi Carlos,

Are you predicting glare using the DGI? if you are, there may be some errors produced by potential glare sources behind the field of view.

findglare identifies potential glare sources. then glarendx -t dgi takes these sources and calculates glare indices in a range of directions. take a look at the predicted dgi's at adjacent directions. if there is an obvious jump in dgi between adjacent directions, then you probably have a glare source behind the field of view. To fix, simply remove the offending glare source from the file output by findglare, and re-run glarendx -t dgi. if that doesn't make any sense, let me know and i'll send you a spreadsheet i developed to identify problematic glare sources.

For Greg, or whoever else might be interested, the problem is in the calculation of dgi's omega in glarendx.c. where the glare source is behind the field of view, posindex returns -1, p becomes 0 and omega is much bigger than adjacent levels. this means the glare source is artificially inflated, and the glare index jumps up.

Phil.

_______________________________
Phillip Greenup
Specialist Engineer
Arup Australasia
Level 10, 201 Kent St
Sydney NSW 2000
Ph (02) 9320 9426
Fax (02) 9320 9321

[email protected] 18/11/2003 9:15:11 pm >>>

Hi Greg,

The pictures were generated using the fish-eye view, they
are looking from the middle of one of the long walls of the
room to the other (the expected position of the user). So I
have the window on one side of the picture and the bottom
of the room on the other.

I give the findglare both the octree and the picture name,
but I get the same results just as if specifying only the
picture. By the way, just giving findglare the octree and
direction of the view gives an error message in DR 2.0. I
have also increased the number of ambient bounces, but it
doesn't change, either.

Carlos

_________________________________________________________
Do You Yahoo!?
Informaci�n de Estados Unidos y Am�rica Latina, en Yahoo! Noticias.
Vis�tanos en http://noticias.espanol.yahoo.com
_______________________________________________