Hi,

I was recently asked by a friend to help him getting started in the use of
his freshly buyed computer. He's a lighting professionnal and wants to take
a chance to give up paying lots of money to badly tempered professionnals
who render his lighting designs for him. He wants also to use the power of
the computer to figure out things for himself.
I'm totally innocent in ray-tracing, but being a programmer, I liked the
pov-ray approach, and I definitely think that the best things are for free.
So I downloaded the renderer, and now I wonder what possibilities exist to
simulate professional lighting designs to be fed to POV. I imagine my friend
to be willing to include special lamps into the scenes he plans to create.
Are such lamps easy to model ?

Thanks for your help,

Candide Kemmler

Post a reply to this message

21th century fox wrote:
> 
> Hi,
> 
> I was recently asked by a friend to help him getting started in the use of
> his freshly buyed computer. He's a lighting professionnal and wants to take
> a chance to give up paying lots of money to badly tempered professionnals
> who render his lighting designs for him. He wants also to use the power of
> the computer to figure out things for himself.
> I'm totally innocent in ray-tracing, but being a programmer, I liked the
> pov-ray approach, and I definitely think that the best things are for free.
> So I downloaded the renderer, and now I wonder what possibilities exist to
> simulate professional lighting designs to be fed to POV. I imagine my friend
> to be willing to include special lamps into the scenes he plans to create.
> Are such lamps easy to model ?
> 
> Thanks for your help,
> 
> Candide Kemmler

I will comment on this but don't think of me as the definitive answer.

Advantages:

  There are several aspects about pov's lighting model that can aid a designer
in his work. Say for instance you place a light source inside of a box with
a certain recess. The designer wants to know were the shadow lines are going
to be on the floor/desk/wall as the light exits the box. Pov could do this
and quite well I would think. It could also be used to find overlapping areas
of illumination from multiple sources. Pov has many different lighting models
to choose from so there are a lot of options he may utilize to meet his goals.
  The problems he will encounter with the above scenarios will be not so much
the intensities of the light but rather the interaction of the light with the
objects it will come in contact with. This is not insurmountable and with a
little practice he should be able to specify default finishes and colors to
suit many "average" situations.

There are many other possible scenarios and I'm hesitant to write a novel here
in reply :)

Disadvantages:

  One major disadvantage of the Pov lighting model is that is does not accurately
reflect the true color temperature of a light source. Fluorescent lights are a
good example of this limitation and is clearly visible if you have ever tried
to use fluorescent lighting in real world photography. The light emitted by a
fluorescent lamp is
full of gaps in the color spectrum. It has some red, skips
12 bands, has some green, skips 30 bands, has some blue (quit a bit really),
skip some more...
  Well you get the picture. This is near impossible to accurately model in pov
and the results will reflect on how it interacts with objects in the scene.
As a result of this I think it would be difficult to use Pov to design the
optimum number of fluorescent light to use in a room for whatever the desired
lighting requirement is.
 More about the color of light is it's temperature rating. Sunlight has a color
temperature
measured in Kelvin's of about 5500 k. Light bulbs have a color
temperature of around 3300 k, some halides and halogens run appox 4500 k. Not
only do these common light sources have different color temperatures they also
put out large amounts of color restricted to a narrow band.
  Taking the above comments into context you can see that it would be difficult
to use pov to model light in a physics correct form of research. On the other
hand you can force Pov to conform to many of the common characteristics of light's
behaviour.
  For example the fade distance fade power light modifiers will allow you to
accurately model the inverse square law. This is the law that explains how much
illumination one
can expect from a uniform light source as it travels a certain
distance from the source. This will allow him to model a light bulb of a certain
intensity and
calculate how bright it should appear as it comes into contact with
an object, say at the other side of a room or area, and make adjustments to it's
placement optimization for lighting a given area.

  In conclusion the answer is yes and no. It will allow him to do many things
but it has it's limitations in the real world physics of light. He may also
find that while the power is there it is both time consuming and tedious to
model a large scale room full of objects, furniture, room dividers, planters,
windows, etc...
  If he is devoted to trying Pov for this purpose, time spent making libraries
of common objects, that are modular in size and design will prove invaluable.
These libraries of objects can be used over and over again adjusting only the
object placement, wall locations, and light sources as needed to prove a
particular design, and his time spent at the computer may be shortened greatly.


-- 
Ken Tyler

mailto://tylereng@pacbell.net

Post a reply to this message

I'd say that if you are concerned with "accurate" light
spec's and acurate specification of surfeace material you
need to look at the radiance package. They have gone to
great lenghts to incorperate reality. See:

	http://radsite.lbl.gov/radiance/HOME.html


21th century fox wrote:
> 
> Hi,
> 
> I was recently asked by a friend to help him getting started in the use of
> his freshly buyed computer. He's a lighting professionnal and wants to take
> a chance to give up paying lots of money to badly tempered professionnals
> who render his lighting designs for him. He wants also to use the power of
> the computer to figure out things for himself.
> I'm totally innocent in ray-tracing, but being a programmer, I liked the
> pov-ray approach, and I definitely think that the best things are for free.
> So I downloaded the renderer, and now I wonder what possibilities exist to
> simulate professional lighting designs to be fed to POV. I imagine my friend
> to be willing to include special lamps into the scenes he plans to create.
> Are such lamps easy to model ?
> 
> Thanks for your help,
> 
> Candide Kemmler

Post a reply to this message

Ken wrote:

>   One major disadvantage of the Pov lighting model is that is does not accurately
> reflect the true color temperature of a light source. Fluorescent lights are a
> good example of this limitation and is clearly visible if you have ever tried
> to use fluorescent lighting in real world photography. The light emitted by a
fluorescent lamp is
> full of gaps in the color spectrum. It has some red, skips
> 12 bands, has some green, skips 30 bands, has some blue (quit a bit really),
> skip some more...

This problem might be worked around by taking into account that povray
does not care about the real color - it just does 3 components with each
ray. One might sample the spectra of the lamps at N (sufficiently
densely
spaced) wavelengths, and run povray N/3 times, giving at each run the
lamp intensities resp. object ref[lr][ea]ctivities for the three
wavelengths used in this run, and finally mix the N/3 output images
(using e.g. ppmarith(1)) using the r/g/b sensitivity curves of the human
eye. Of course, this approach is awkward.

Using only 3 color channels, no program would be able to 
show the lamp-dependent impression of objects with metameric
colors.

I guess the main problem for real world lighting applications
might be the radiosity effects, which (in places properly lit
for humans) dominate. I don't know how perfect the radiosity
implementation in povray is.

Ralf

Post a reply to this message