[Radiance-general] Understanding illum correctly?

[Gregory J. Ward]

Hi Mark,

I realize this is confusing, but basically you want to put a "void"  
polygon in front of your opening.  Carsten is correct that the  
polygon needs to point into the room, but some other slight  
modifications are necessary as well to work with rad, as identified  
below.  It is probably better just to use a pane of glass instead of  
an illum initially.

> From: Mark Shewfelt <MShewfelt at enermodal.com>
> Date: December 7, 2006 1:25:21 PM PST
>
> Hello All,
>
> I hoping someone can help me with my understanding of illum sources.
> I've read RwR, done the tutorials, etc., and I think I'm OK with the
> concept of secondary sources. All of the examples I've seen make  
> illums
> around windows and skylights. My situation is slightly different - I
> have a skylight that is not flush with the interior of the space (i.e.
> there is a well, the insides of which are painted white).
>
>
>   -----------   <-- skylight glass
>   |         |
>   |         |   <-- well
>   |         |
>   |         |
> --	 __    ---------------  <-- ceiling of room
>         \
>          \
> 	    -------------------- illum
>
>    <- room ->
>
>
> My intention is to make an illum at the ceiling of the room, which
> incorporates the bounces that the light takes to get into the room.
> Using the files below I've been able to simulate the skylight and the
> results look pretty reasonable. However, when I run the simulation  
> using
> rad, I get the error message 'warning - no light sources found'. From
> reading Axel Jacobs' excellent tutorials I was imagining that this
> message would go away with the illums.
>
> Any comments/tips/pointers would be great.
>
> Cheers,
>
> Mark Shewfelt
> Enermodal Engineering Ltd.
> Kitchener Ontario Canada
>
>
> skylight.rif:
> ------------------------------
> scene = sky.rad
> scene = material.mat
> scene = room3.rad
> illum = skylight.rad
> # The file in which to store indirect values
> AMB= simple.amb
> INDIRECT = 2
> mkillum= -ab 2
>
>
> material.mat:
> ------------------------------
> skyfunc brightfunc window_dist
> 2 winxmit winxmit.cal
> 0
> 0
>
> window_dist glass skylight_glass
> 0
> 0
> 3	0.8	0.8	0.8

Replace the above with simple, clear glass type:

void glass skylight_glass
0
0
3 0.96 0.96 0.96

> # light is emitted from this but is invisible.
> void illum skylight_illum
> 0
> 0
> 3	0	0	0

You can take the illum out, as we won't need it (and the zeroes  
shouldn't be there, as Carsten mentioned).

> skylight.rad
> ------------------------------
> # light enters skylight well from here
> skylight_glass polygon skylight_window
> 0
> 0
> 12
>  	5.491076	 6.503552	 3.400000
>
> 	 4.201076	 6.503552	 3.400000
>
> 	 4.201076	 5.213552	 3.400000
>
> 	 5.491076	 5.213552	 3.400000

Take the above and put it into room3.rad, as you won't be making it  
into an illum with mkillum.  The "illum=" line in your rad input file  
is for surfaces to be made into illum's, only.  Orientation doesn't  
matter in this case.

> # light is emitted from this "surface" into the room
> skylight_illum polygon face.65
> 0
> 0
> 12
> 	 5.491076	 6.503552	 3.00000
>
> 	 4.201076	 6.503552	 3.00000
>
> 	 4.201076	 5.213552	 3.00000
>
> 	 5.491076	 5.213552	 3.00000

As Carsten points out, you need to reverse your vertex order:

# light is emitted from this "surface" into the room
void polygon face.65
0
0
12
	 5.491076	 5.213552	 3.00000
	 4.201076	 5.213552	 3.00000
	 4.201076	 6.503552	 3.00000
	 5.491076	 6.503552	 3.00000

---------------
Notice that I replaced the modifier with "void."  Mkillum will take  
this file and compute an appropriate output distribution based on the  
light entering the space and up to 2 indirect bounces.  (This may be  
modified by setting the "mkillum=" variable in rad.)

-Greg