Hello all,

I'm working a theatrical lighting system library for povray, based on LightSysIV
and CIE by Jaime and Ive, and I'm running into two problems with spotlights.

The first is trying to simulate a PARCan
<http://en.wikipedia.org/wiki/Parabolic_aluminized_reflector_light>, which is
an instrument with a lens that produces a soft-edged oval shape.
I've already implemented the standard spotlight with the built in function (with
beam and field angle), but can't figure this out without creating a complex
glass-textured asymmetrical solid lens. Does anyone know how to create this
effect?

The second is a problem with the color of the falloff of the standard spotlight.
If I use a light color that is slightly off-white (to be exact, using
Blackbody(3200) from CIE), the center of the beam is the correct color but it
changes from light yellow-ish to a deep orange between the beam radius and
falloff radius. This is projecting onto a pure white surface (diffuse 1 ambient
0), and using light fading. I didn't specify the fade_color, since I assume it
is black? How can I get rid of this effect, as it doesn't model real light
behavior at all? (possibly helpful addition: the original light color is way
over rgb 1.0 (closer to rgb 1*10000), to allow for a distant light)

Thanks for any help!

Post a reply to this message

cshake wrote:

> behavior at all? (possibly helpful addition: the original light color is way
> over rgb 1.0 (closer to rgb 1*10000), to allow for a distant light)

Sounds like a clipping effect. This would means your high intensity
color actually speficies the orange, not the yellow, but in the high
intensity areas the red channel is oversaturated and clipped to the
maximum supported by the output format (typically 8 bit), so the
ratio of red/green becomes smaller and thus more yellow-ish.

Try reducing the light intensity to observe the change.

Post a reply to this message

Christian Froeschlin wrote:
> cshake wrote:
> 
>> behavior at all? (possibly helpful addition: the original light color 
>> is way
>> over rgb 1.0 (closer to rgb 1*10000), to allow for a distant light)
> 
> Sounds like a clipping effect. This would means your high intensity
> color actually speficies the orange, not the yellow, but in the high
> intensity areas the red channel is oversaturated and clipped to the
> maximum supported by the output format (typically 8 bit), so the
> ratio of red/green becomes smaller and thus more yellow-ish.
> 
> Try reducing the light intensity to observe the change.

Thanks, reducing the intensity did help, I guess I was expecting the 
color to be closer to white than it really was. I also didn't realize 
that clipping or oversaturation could be an issue, had never run into it 
before.
I rendered the test scene again and found that increasing the bit depth 
of the output file from 8 to 16 (png) didn't make a difference itself, 
but either dimming the light or moving it farther away (with fading) did 
do the trick.

I'm still stuck on the other issue, but this gets me somewhere.

Post a reply to this message

Don't know if this will be completely right for what you want to do, using 
projected_through might be one solution.
Render the following to see this "oval" spotlight shine onto a wall and 
floor, no camera needed. You might need to reposition it to see it better, 
just notice this is inside a box if you do.
There could be problems trying to adjust the actual light to fit your needs. 
Not sure if area_light likes projected_through, or vice versa.

/* BEGIN */
light_source
{ // placed at origin, pointing toward +z
    0,
    1  // color
    spotlight
    point_at z
    radius 5
    falloff 25
    area_light x/30,y/15,9,18 adaptive 1 jitter
    projected_through
    {
        sphere
        { // oval object placed in front of spotlight
            z, 0.5
            scale <1/3,1,1>/6
        }
    }
    rotate 45*x // angle downward
    translate 6*y // position upward
}

box
{ // floor and wall, for light to shine onto
    <-3,-1,-3>,<3,9,9>
    pigment { rgb 1 }
    finish { ambient 0 diffuse 1 }
}
/* END */

Post a reply to this message

CShake wrote:

> I rendered the test scene again and found that increasing the bit depth 
> of the output file from 8 to 16 (png) didn't make a difference itself

yes, I was confused about that one. You still get the same intensity
range with 16 bit but with finer resolution. It may still be useful for
you because it means you can use very low intensity light sources to
avoid the clipping effect and still get a high dynamic range.

Post a reply to this message

cshake nous illumina en ce 2009-02-15 00:10 -->
> Hello all,
> 
> I'm working a theatrical lighting system library for povray, based on LightSysIV
> and CIE by Jaime and Ive, and I'm running into two problems with spotlights.
> 
> The first is trying to simulate a PARCan
> <http://en.wikipedia.org/wiki/Parabolic_aluminized_reflector_light>, which is
> an instrument with a lens that produces a soft-edged oval shape.
> I've already implemented the standard spotlight with the built in function (with
> beam and field angle), but can't figure this out without creating a complex
> glass-textured asymmetrical solid lens. Does anyone know how to create this
> effect?
> 
> The second is a problem with the color of the falloff of the standard spotlight.
> If I use a light color that is slightly off-white (to be exact, using
> Blackbody(3200) from CIE), the center of the beam is the correct color but it
> changes from light yellow-ish to a deep orange between the beam radius and
> falloff radius. This is projecting onto a pure white surface (diffuse 1 ambient
> 0), and using light fading. I didn't specify the fade_color, since I assume it
> is black? How can I get rid of this effect, as it doesn't model real light
> behavior at all? (possibly helpful addition: the original light color is way
> over rgb 1.0 (closer to rgb 1*10000), to allow for a distant light)
> 
> Thanks for any help!
> 
> 
Blackbody(3200) is rather redish, not slightly off-white, similar to a spotlight 
running with a much lower voltage than it's rating.
A value around 4500 to 5000 would give you a much yellower tint. And around 6500 
will be close to white. 7000 will give a marked blue cast.

If you see the cender of the beam in a weak reflector, it will look prety orange.

fade_color is NOT used for lights, but for coloured transparent materials and is 
found in the interior block of such objects.

-- 
Alain
-------------------------------------------------
Fundamentalism: If shit happens to a televangelist, it's okay.

Post a reply to this message

Alain wrote:
> Blackbody(3200) is rather redish, not slightly off-white, similar to a 
> spotlight running with a much lower voltage than it's rating.
> A value around 4500 to 5000 would give you a much yellower tint. And 
> around 6500 will be close to white. 7000 will give a marked blue cast.
> 
> If you see the cender of the beam in a weak reflector, it will look 
> prety orange.
> 
> fade_color is NOT used for lights, but for coloured transparent 
> materials and is found in the interior block of such objects.
> 

Thanks for the clarification on fade_color, I saw it in the help file in 
the same section as spotlight so I was unsure.

The reason I'm using a low temperature blackbody is that the lamp I'm 
trying to simulate (the HPL740 used in ETC brand source four lights) is 
at that temperature, and most other stage light instruments (except arc 
lamps) are around 3200K. The color gels I'm simulating were made for use 
with this temperature light, so for most light sources there is also a 
thin object with a color filter value in front of it.

Post a reply to this message

CShake escribió:
> Alain wrote:
>> Blackbody(3200) is rather redish, not slightly off-white, similar to a 
>> spotlight running with a much lower voltage than it's rating.
>> A value around 4500 to 5000 would give you a much yellower tint. And 
>> around 6500 will be close to white. 7000 will give a marked blue cast.
>>
> The reason I'm using a low temperature blackbody is that the lamp I'm 
> trying to simulate (the HPL740 used in ETC brand source four lights) is 
> at that temperature, and most other stage light instruments (except arc 
> lamps) are around 3200K. The color gels I'm simulating were made for use 
> with this temperature light, so for most light sources there is also a 
> thin object with a color filter value in front of it.

   Remember you can always change the color system white point to something 
more like an indoor film. Here are some examples from the Lightsys demos:

// Change color system white point for lights
//CIE_ColorSystemWhitepoint(My_ColSys, Illuminant_A)    // tungsten 2856K
//CIE_ColorSystemWhitepoint(My_ColSys, Blackbody2Whitepoint(3200)) // Film 
tungsten A
CIE_ColorSystemWhitepoint(My_ColSys, Blackbody2Whitepoint(3400)) // Film 
tungsten B

   I think a 3400K white point should make the 3200K light to look slight 
yellowish.

   Regards,

--
Jaime

Post a reply to this message