Basic daylighting and artificial lighting test(ambient value problem)

Hi Nalin.

Welcome to the wonderfully complex world of Radiance.

My answers are below ...

Hello all

I am new to radiance and am trying to learn its various features
I have modeled a room 16*6*3 in ESP-r and then exported the model to
radiance now I am trying to simulate this for two times of a day one at 1200
noon(clear sky) and other at 20:00(night)

I'm not very familiar with ESP-r and how it models scenes and converts
to Radiance. You should check that 20:00h is actually dark for the day
you selected. Depending on your location and time of the year the sun
can still be visible.

Also I assume that ESP-r models walls as abstract 'interfaces' with
certain properties. These get exported to Radiance as single polygons.
The light leaks you see in your image are effects of interpolation, where
Radiance incorrectly calculates the brighness of the wall based on the
outside illuminance. To avoid this you can switch interpolation off (big
impact on rendering time, option '-aa 0'). More efficient would be to
build the model with 'real' walls, so that your interior 'walls' are boxed
inside the exterior 'walls'. That needs to be supported by your exporter,

what I want to ask is what should be my basic rendering settings to get a
decent enough output and what should I set the ambient value(-av) to for the
night case should it be zero?

The ambient value is necessary to compensate for the loss of
illuminance that Radiance cuts off when it limits the number of bounces.
Basically every further bounce would still contribute to the illuminance
but if you stop following the ray after N bounces you neglect this
small amount. Setting the '-av' value to the average of the room will
then add this amount. You could say, with the right '-av' value you
get a more accurate result than without.

Finding the right '-av' value is an art in itself. For outside scenes
you get a value in the comments of the output of the gensky program.
Run gensky in a terminal ('command prompt') with your date and time
and look for the comment line '# Ground ambient level:'.

For indoor scene (without artificial light) you can use this value and
multiply it with a reasonable daylight factor for the room.

If you add aritificial light you can calculate an approximate '-av'
from your target horizontal illuminance (in Watts/m^2) divided
by Pi. Say you want 200lux (0.29W/m^2) then your -av would
be 0.29/3.14 = 0.092.

I want to check the illuminance value for the
room during the day and using artificial lights during the night set the
illuminance level of the room to say 200 lux but the problem is if I give
some value to the ambient value(-av) it adds up in the lumen value of only
artificial light and gives higher output so for both daylighting and
artificial lighting at every part in the room I am getting high lumen
values, so, is this correct?should ambient value ever be used for realistic
results? also for the daylight analysis what should be the ambient value and

I think I answerd all the above. I might have missed a point here
but that's due to your lack of interpunctuation. Capitalisation is power!

also does the illumination value depend on the no. of bounces
if yes then how would I know which no. of bounces to give for correct

That's another thing where you need experience and you have
to define how accurate you want your renderings to be. Your
scene (based on the image) is empty so most light hits the
floor and walls directly from the luminaires (assuming there is
no contribution from the outside). You have to make a guess
about the number of significant bounces. In this situation I'd
say you are reasonably accurate after 3 or 4 bounces.

To verify your estimate you can calculate a single point with rtrace
and increase the number of bounces until your results do not
vary significantly.

Also don't expect Radiance to be unrealistically accurate. I always
got frustrated by Lighting Designers who believe that Radiance can
give you the lux values or daylight factors to the second decimal place.
Even your average light meter has an error of 3%. They have to learn
how to deal with good estimates.



On Sun, Jul 12, 2009 at 8:33 AM, nalin swaroop<[email protected]> wrote:

Hi Thomas, what if a small diameter radiance "cylinder" could be drawn
around the edges with the light leaks? Like a kind of "silicone sealant"!
As you know, I'm no coder but I'm guessing it could be python scripted from

Interesting idea. I don't know if this would have the desired effect, though,
or just introduce more interpolation errors.

There'd be a small increase in rendering time. If the cylinder is
small enough it won't show. Or, if the appropriate size and colour are
selected you could achieve "cartoon edges"!

As intended graphical effect this is a nice way to do it. But if you
export simple scenes from the likes of ESP-r it should also be
possible to work out a wall thickness from the attributes of the wall.
If only a single room is exported it should be quit easy to apply an
offset to the floor outline and create two layers of polygons. Usually
the geometries in energy simulation apps are quite basic which
simplifies the calculation of intersections and openings.

But for the moment those are still features for the next generation
of exporters.



On Tue, Jul 14, 2009 at 11:38 AM, Chris Yates<[email protected]> wrote: