Radiance simulations maintain 200 lux minimum

Hi Caroline,

Thanks for posting your input and images. The mostly likely cause of your problems is that you left rad to set the default value for -av, which is based on your exposure. Since you use the -2 option to pfilt, you don't see the effect this exposure setting has on your image, therefore you don't realize that it is probably too high. You need to either manually set the -av value in your render= line, or figure out what the correct exposure should be for your scene and use that in your EXPOSURE= setting. Then, the light should be better balanced.

A lot of people set -av 0 0 0 as a matter of course in their daylighting simulations. Although such a conservative setting is guaranteed to underestimate your lighting levels, it is often better than ending up with too much light where it does not belong, like you are seeing here. At least you noticed it!

The Unix version of radiance comes with a script called compamb that automatically determines an appropriate ambient value (and exposure) given a rad input file. I don't believe this is included in DesktopRadiance, but you can look for it.

Good luck!

Hi Caroline,

I can think of three possible things that you might want to look into:

   1. Illums - these are imposter geometry to help optimize daylight
      simulation in interior environments. I see that you have the
      illum= line in the rif file, however since you are getting the
      warning about "no lights found" it is likely that there is some
      problem: A) illum geometry is directionally dependant, the surface
      normal must point into the room. To create a polygon in Autocad
      with the normal pointing into the room, the polygon (a closed
      pline or 3dface for example) can be constructed by generating
      vertices in a counter clockwise manner, for example if the polygon
      is built in the xy plane then drawing it counter clockwise would
      result in a normal that point in positive z. B) I do not use the
      rad program so I am not so sure about this, but does the geometry
      that appears in the illum= line also need to apear in the scene= line?
   2. Ambient value - this allows a baseline ambient value to be
      established for the scene, this can be estimated or precalculated
      with a different script (although I am not sure how DR deals with
      this). If this value is too high then the images can have the kind
      of "glowy" appearance. I am guessing that a default value is being
      used from the information in the rif file. This may need to
      estimated by hand for best results.
   3. Indirect - this indicates how many ambient bounces to calculate.
      If you are after accurate results this number probably needs to be
      2 and perhaps more depending on other factors.


Caroline Prochazka wrote:


Hello all,

As Greg suggested, I have posted my results to my
website. No fancy links between them, and the text
files are done as jpegs, but the info is there.

I ran simulations of a 15m x 15m room. The west wall
has 4 windows and the north, south and east walls each
have 1 central window. I simulated December 21st in
Toronto for noon, 4pm and 6pm. According to other
software and my weather data, there is no sunlight at
6pm on Dec21, and the sun is quite low, but not yet
set at 4pm. I did not model any exterior obstructions.

Using the Desktop Radiance interface in AutoCAD I
generated .rif files for the time/day selections
(Radiance>Simulation>Camera...). Then in Command
Prompt I performed RAD on the various .rif files.

The sky definitions all appear identical, except for
the time markers in the top line.
The .riff and sky.rad files are posted here:

The RAD command ran through my camera views
successfully, but generated some warnings:
warning - no light source found. Well of course, I
have no luminaires (as suggested in the Desktop
Radiance FAQ), I am studying daylighting, so I presume
I can move on despite this message.
warning - aiming failure for light source
"softwaretest4_38.4.2". Looking at the DesktopRadiance
FAQ, online, there would not seem to be any reason for
this warning, based on the model I have drawn. No
concave items, no cylinders, no rings with holes, etc.

The entire script for my RAD run of Dec 21 at noon is
posted here:
apologies for the fuzziness.

So the resulting images are posted as follows:
December 21, noon:
December 21, 4pm:
December 21, 6pm:
The false color maps are generated within the winimage
screen (Analysis>False color>choose metric,
illuminance, max=1000, intervals=10)

Note the centre of the floor of the room (roughly the
centre of the floor area as I have cast my camera
view) is near 200 lux in all three cases, even when
there are no luminaires and the scene is after

I've also noticed that the 6pm version produces an
ambient glow in the room, both in general and more
specifically along the ceiling edge. Where is this
light coming from?

Things I HAVE checked. 1.All planes are in contact along all edges. 2.All windows have their surface normals pointing inwards. 3.Double and triple checked (multiple runs with same
results) each of these simulations.

I am a new user so my use of the Command Prompt to do
most of this is limited. Any suggestions would be
greatly appreciated, though.


Do you Yahoo!?
Yahoo! Photos: High-quality 4x6 digital prints for 25�

Radiance-general mailing list
[email protected]

# John E. de Valpine
# president
# visarc incorporated
# http://www.visarc.com
# channeling technology for superior design and construction

Jack de Valpine wrote:

      B) I do not use the
      rad program so I am not so sure about this, but does the geometry
      that appears in the illum= line also need to apear in the scene= line?




      Rob Guglielmetti

e. [email protected]
w. www.rumblestrip.org

Great! Thanks for the suggestions. Indeed it seems as
though -av will fix this. Next step is to try this on
my larger simulations.
An additional questions, as I am not too experienced
with this application:

As my primary use of Desktop Radiance centres on the
changes in daylighting in a building as the glazing
type changes, should I expect to need a different -av
setting for each variation of the design? What is
likely to be the greatest factor in setting my -av
number - does it sound reasonable that for the same
buidling geometry, I will probably find the same -av
setting works throughout all glazing variations? Or
will it be the date/time/location setting which is the
biggest player? or something else?

...or should I just plan for this to take several
tests of -av settings for each and every glazing



Do you Yahoo!?
Yahoo! Photos: High-quality 4x6 digital prints for 25�

Hi Caroline,

If you are changing all the glazing at once, you can simply scale the -av parameter value by the transmittance of your glazings and that will give you the correct relative changes. In fact, excepting for angular variations in transmittance, you could scale your entire result under those circumstances.