clipka <ano### [at] anonymousorg> wrote:
> Am 18.02.2011 11:03, schrieb Bruno Cabasson:
>
> > I also set the fade_distance in the same order of magnitude than the area light
> > (twice). I am a bit confused by the fact that the dome is so barely visible,
> > despite 4 candles. But, after all, they are small (the biggest is 16 cm high)
> > and the dome is 5 metre from there. Maybe relative illuminations are correct.
>
> I never liked that dome anyway :-P

Nor did I... It was just a basic checkered texture (dome was not the purpose of
the scene). Floor was plain white for tweaking lighting/shadows/radiosity. I'll
have another go with something more acceptable. Thinking of some igloo with SSLT
snow blocks, why not ...

Post a reply to this message

> "Edouard"<pov### [at] edouardinfo>  wrote:
>>> Is this one better?
>>
>> The wax looks just right, you can see the sizes just at a glance by the SSLT!
>>
>> Cheers,
>> Edouard.
>
> In this experiment, I intended to make the size of the flame proportional to the
> size of the candle. So, I adjusted the lights powers according to the sizes
> proportionally, ie: *2 for the biggest, *1, *1/2, *1/4 respectively. I don't
> know the law for the illumination of flames wrt their size. I suppose the
> illumination comes from the surface of the flame, and therefore follows a
> quadratic law. My powers are surely wrong for flame powers proportional to the
> flame sizes.
>
> However, in RL, the size of the flame is generally not proportional to the size
> of the candle ...
>
> I also set the fade_distance in the same order of magnitude than the area light
> (twice). I am a bit confused by the fact that the dome is so barely visible,
> despite 4 candles. But, after all, they are small (the biggest is 16 cm high)
> and the dome is 5 metre from there. Maybe relative illuminations are correct.
>

In a flame, the light comes from the whole volume, and the flame is 
essentialy transparent.

Make this test: light a candle and light some other light, place them to 
project the shadow on a wall or similar surface. Now, take a look of the 
shadow of the candle and it's flame. It's only effect on the light 
passing through it is from it's lower than air ior.

Even the flame of a camp fire is essentialy transparent. You see it only 
because the gaz of the flame is incandescent and emit light.
Same thing with the flame from a gaz stove.
Model those flames using emissive media, without absorbtion nor 
scattering media.




Alain

Post a reply to this message

Alain <aze### [at] qwertyorg> wrote:

> > "Edouard"<pov### [at] edouardinfo>  wrote:
> >>> Is this one better?
> >>
> >> The wax looks just right, you can see the sizes just at a glance by the SSLT!
> >>
> >> Cheers,
> >> Edouard.
> >
> > In this experiment, I intended to make the size of the flame proportional to the
> > size of the candle. So, I adjusted the lights powers according to the sizes
> > proportionally, ie: *2 for the biggest, *1, *1/2, *1/4 respectively. I don't
> > know the law for the illumination of flames wrt their size. I suppose the
> > illumination comes from the surface of the flame, and therefore follows a
> > quadratic law. My powers are surely wrong for flame powers proportional to the
> > flame sizes.
> >
> > However, in RL, the size of the flame is generally not proportional to the size
> > of the candle ...
> >
> > I also set the fade_distance in the same order of magnitude than the area light
> > (twice). I am a bit confused by the fact that the dome is so barely visible,
> > despite 4 candles. But, after all, they are small (the biggest is 16 cm high)
> > and the dome is 5 metre from there. Maybe relative illuminations are correct.
> >
>
> In a flame, the light comes from the whole volume, and the flame is
> essentialy transparent.
>
> Make this test: light a candle and light some other light, place them to
> project the shadow on a wall or similar surface. Now, take a look of the
> shadow of the candle and it's flame. It's only effect on the light
> passing through it is from it's lower than air ior.
>
> Even the flame of a camp fire is essentialy transparent. You see it only
> because the gaz of the flame is incandescent and emit light.
> Same thing with the flame from a gaz stove.
> Model those flames using emissive media, without absorbtion nor
> scattering media.
>
>
>
>
> Alain

Thanks for advise.

Do you have such a model you can share?

I think however that the incandescent particles in the flame scatter the light
of the others, and that they are also somewhat opaque and have an absorbing
behaviour. Other particles or atoms might also scatter their light. It seems
that a flame is quite a complex object.

Post a reply to this message

Am 20.02.2011 22:59, schrieb Bruno Cabasson:

> I think however that the incandescent particles in the flame scatter the light
> of the others, and that they are also somewhat opaque and have an absorbing
> behaviour. Other particles or atoms might also scatter their light. It seems
> that a flame is quite a complex object.

That is indeed the case for sooty flames. A cleanly burning candle flame 
won't show much of that effect though.

Post a reply to this message

> Alain<aze### [at] qwertyorg>  wrote:

>>> "Edouard"<pov### [at] edouardinfo>   wrote:
>>>>> Is this one better?
>>>>
>>>> The wax looks just right, you can see the sizes just at a glance by the SSLT!
>>>>
>>>> Cheers,
>>>> Edouard.
>>>
>>> In this experiment, I intended to make the size of the flame proportional to the
>>> size of the candle. So, I adjusted the lights powers according to the sizes
>>> proportionally, ie: *2 for the biggest, *1, *1/2, *1/4 respectively. I don't
>>> know the law for the illumination of flames wrt their size. I suppose the
>>> illumination comes from the surface of the flame, and therefore follows a
>>> quadratic law. My powers are surely wrong for flame powers proportional to the
>>> flame sizes.
>>>
>>> However, in RL, the size of the flame is generally not proportional to the size
>>> of the candle ...
>>>
>>> I also set the fade_distance in the same order of magnitude than the area light
>>> (twice). I am a bit confused by the fact that the dome is so barely visible,
>>> despite 4 candles. But, after all, they are small (the biggest is 16 cm high)
>>> and the dome is 5 metre from there. Maybe relative illuminations are correct.
>>>
>>
>> In a flame, the light comes from the whole volume, and the flame is
>> essentialy transparent.
>>
>> Make this test: light a candle and light some other light, place them to
>> project the shadow on a wall or similar surface. Now, take a look of the
>> shadow of the candle and it's flame. It's only effect on the light
>> passing through it is from it's lower than air ior.
>>
>> Even the flame of a camp fire is essentialy transparent. You see it only
>> because the gaz of the flame is incandescent and emit light.
>> Same thing with the flame from a gaz stove.
>> Model those flames using emissive media, without absorbtion nor
>> scattering media.
>>
>>
>>
>>
>> Alain
>
> Thanks for advise.
>
> Do you have such a model you can share?
>
> I think however that the incandescent particles in the flame scatter the light
> of the others, and that they are also somewhat opaque and have an absorbing
> behaviour. Other particles or atoms might also scatter their light. It seems
> that a flame is quite a complex object.
>
>

Siory, I don't have a ready made flame model at the moment.

In a clean flame, the particles are the same, or about the same, 
dimentions as the molecules from the surounding air. They are small 
molecule sized. So, the flame is about as "opaque" as the air itself.

Sample: candle flame, alchool flame, Bunsen burner, gaz stove burner, 
soldering torch, plasma torch and cutters, controled light oil flame, 
and any oxygen rich flame.

In a camp fire, the top of the flame can contain some sooth and smoke. 
The bottom parts are quite transparent.




Alain

Post a reply to this message

If you want a candle flame, here is one I developed a number of years ago based
on many photos of flames, and had posted here somewhere before.
I always found that people tend to not get the colours right, especially the
purple and blues near the bottom.  This is purely emissive media and has
turbulence added in.  (the turb_translate macro within the macro is to back
check the flame base point so it is still centred where it should be. don't set
the turbulence level too high or it can fail at finding the point and
get stuck in an infinite loop, the sample image is set to 0.3).

sample usage:
union{
    cylinder{0 y*100 10 pigment{rgb 1}}
    cylinder{0 y*106 0.5 pigment{rgb 0}}
    object{Flam(0.3) scale 15 translate y*(108)}
    light_source{0
      rgb <1,.5,0>
      translate <0,117,0>
      media_attenuation on
      fade_power 2
      fade_distance 100
    }
}



//START

#macro Flam (Turb)
  #local R1=seed(2);
  #local OS=rand(R1)*50;
  #local Flam1 = function(x,y,z)
{pow(pow(pow(3*x,2)+pow(0.8*y-0.2,2)+pow(3*z,2),3)+pow(max(1-sqrt(pow(5*x,2)+pow(y+.5,2)+pow(5*z,2)),0),2))}
  #local Flam2 = function(x,y,z)
{floor(pow(sqrt(pow((3.5*x),2)+pow((0.4*y+0.6),2)+pow((3.5*z),2)),2))}
  #local FlamF = function(x,y,z) {min(max(Flam1(x,y,z),Flam2(x,y,z)),1)}
  #macro turb_translate (tTurb,tOcta,tLamb,tOmeg,P0,Acc)
    #local PC=P0;
    #local Dir=<0,0,0>;
    #local V0=vlength(vturbulence(tLamb, tOmeg, tOcta, P0)*tTurb);
    #local Step=V0/2;
    #macro AB(PC,Step,d,Dir,V0)
      #local P1=PC+Step*d;
      #local V1=vlength(P1+vturbulence(tLamb, tOmeg, tOcta, P1)*tTurb-P0);
      #if (V1<V0 | d.x=1)
        #local Dir=d;
        #local V0=V1;
      #end
    #end
    #while (V0>Acc)
      #local DrP=Dir;
      AB(PC,Step, x,Dir,V0)
      AB(PC,Step,-x,Dir,V0)
      AB(PC,Step, y,Dir,V0)
      AB(PC,Step,-y,Dir,V0)
      AB(PC,Step, z,Dir,V0)
      AB(PC,Step,-z,Dir,V0)
      #if (vlength(DrP+Dir)=0)
        #local Step=Step/2;
      #else
        #local Step=V0/2;
      #end
      #local PC=PC+Step*Dir;
    #end
    PC
  #end
  #local tOcta=7;
  #local tLamb=1.32;
  #local tOmeg=0.35;
  sphere{0 1.1 scale <1/2,1,1/2> translate y*(0.15+OS)
    hollow
    pigment{rgbt <1,1,1,1>}
    interior{
      media{
        emission rgb 0.3
        method 3
        intervals 2
        density{
          function{FlamF(x,y-OS,z)}
          color_map{
            [0.00 rgb <1,1,.5>]
            [0.20 rgb <1,.5,0>]
            [0.60 rgb <1,.1,0>]
            [0.78 rgb <0,0,0>]
            [0.90 rgb <0,0,0>]
            [0.94 rgb <0,0,1>]
            [0.98 rgb <0,0,1>]
            [1.00 rgb <0,0,0>]
          }
          warp{
            turbulence Turb
            octaves tOcta
            lambda tLamb
            omega tOmeg
          }
        }
      }
    }
    no_shadow
    translate turb_translate(Turb,tOcta,tLamb,tOmeg,y*(0.75-OS),0.001)
  }
#end

//END

Post a reply to this message

Attachments:
Download 'candleflame.jpg' (396 KB)

Preview of image 'candleflame.jpg'

Good job!

Here is a more modest code snippet with emission-only media in a simple ovus
container (could be an animatable blob)

#include "colors.inc"

global_settings
{
    assumed_gamma 1
}

#declare FLAME_COLOR1 = (2*Firebrick+Goldenrod)/3;
#declare FLAME_COLOR2 = (2*Orange+Goldenrod)/3;

camera
{
    location -10*z
    angle 40
    right x*image_width/image_height
}

sphere
{
    0, 1
    hollow
    pigment {brick scale 0.5*x scale 1/250 warp {spherical}}
    scale 30
}

ovus
{
    1, 0.6
    pigment {rgbt 1}
    hollow
    interior
    {
        media
        {
            samples 5
            emission 6
            //scattering {1, White}
            //absorption 4
            density
            {
                spherical scale 2
                translate -1.5*y
                color_map
                {
                    [0 FLAME_COLOR1]
                    [0.25 FLAME_COLOR2/8]
                    [0.30 FLAME_COLOR2/8]
                    [0.40 MediumBlue/4]
                    [0.45 MediumBlue/10]
                    [0.6 rgb 0]
                }
            }
        }
    }
    scale 1.5*y
    scale 1/(1.3*1.5)

}

Post a reply to this message

Attachments:
Download 'flame.png' (50 KB)

Preview of image 'flame.png'

Searching the web for informations about flame structure and properties, I found
this paper: http://www.newton.dep.anl.gov/askasci/chem00/chem00873.htm

Post a reply to this message