 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Whilst expirimenting with the media feature, I've noticed something odd.
A media with absorption 1.0 blocks all passing light, given that the way
of the light through the object is 1 unit long. That makes sense.
When adding emission rgb <0,0,1> one would expect the obstacle to appear
uniformly blue.
However, I've noticed that where the background contains more blue, the
intensity of the obstacle is higher too. But the absorption 1.0 should
have blocked all background light, no ?
I've created a small sample scene that shows this. It renders a
background of red, blue and green hexagons, and a box with the described
media.
Comment away the emission value and the object will appear completely
black (except for the edges). With emission, the blue hexgons in the
scene appear brighter through the object.
This makes no sense in my head. Is it just my head or is this something
the POV-team has overlooked ? If all you brains out there don't get it
either, I'd better notify the POV-team.
// ---------------- poser scene ----------------
global_settings { assumed_gamma 2.2 }
#include "colors.inc"
camera {
location <0, 0, -7>
direction <0, 0, 2>
look_at <0,0,0>
}
object {light_source{<0,1000,-1000> color White}}
// hexagon wall
plane { z, 10
pigment { hexagon color Green color Blue color Red rotate <90,0,0>}
hollow
}
box { <-1, -1, -0.5>, <1, 1, 0.5>
hollow
material
{
texture {
pigment {color rgbf 1.0}
finish {diffuse 0 ambient 0}
}
interior
{
media
{
samples 3,10
absorption rgb 1
emission rgb <0,0,1>
/* scattering {0, rgb <0,0,5>
extinction 0.0} */
}
}
}
}
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
H.A. van der Meiden wrote:
>
> Whilst expirimenting with the media feature, I've noticed something odd.
>
> A media with absorption 1.0 blocks all passing light, given that the way
> of the light through the object is 1 unit long.
Actually, this isn't true. There is an exponential fall-off of light
as it passes through the object : in a given distance, a given fraction
of the light is absorbed. But over any finite distance, a finite
fraction of the light survives, subject to ADC bailout and color
quantization.
To see what I mean, increase the absorption to 100. Now it comes
a lot closer to blocking all of the light.
> That makes sense.
> When adding emission rgb <0,0,1> one would expect the obstacle to appear
> uniformly blue.
> However, I've noticed that where the background contains more blue, the
> intensity of the obstacle is higher too. But the absorption 1.0 should
> have blocked all background light, no ?
No. Note, as well, that the media INTERIOR is emitting as well as
absorbing.... so absorption 10 emission 100 will look different
than absorption 100 emission 10 -- in the former the emitter will
dominate while in the latter the absorber will dominate.
> I've created a small sample scene that shows this. It renders a
> background of red, blue and green hexagons, and a box with the described
> media.
> Comment away the emission value and the object will appear completely
> black (except for the edges). With emission, the blue hexgons in the
> scene appear brighter through the object.
This file is an excellent "intro" to media! (I am a relative
newbie at them, myself.) But commenting out the absorption
on my 3.1b6b results in a gray translucence, not black opaqueness.
--
http://www.flash.net/~djconnel/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Dan Connelly wrote:
>
> H.A. van der Meiden wrote:
> >
> > Whilst expirimenting with the media feature, I've noticed something odd.
> >
> > A media with absorption 1.0 blocks all passing light, given that the way
> > of the light through the object is 1 unit long.
>
> Actually, this isn't true. There is an exponential fall-off of light
> as it passes through the object : in a given distance, a given fraction
> of the light is absorbed. But over any finite distance, a finite
> fraction of the light survives, subject to ADC bailout and color
> quantization.
You're right !
After some trial it seems that the absorption parameter is an
exponentional falloff rate.
If r=1 then, at 1 unit distance, exp(-1)=37% of the light remains.
If r=2 then exp(-2)=14% of the light remains.
Basically an absorbtion rate '-log(fraction)/distance' determines that
'fraction' of the light remains after 'distance' units.
Supposedly the emission parameter works in a similar way, determining
some exponentional increase of light.
Unfortunatly high values for both absorption and emission (>2.0) result
in noisy images, although high sample rates seem reduce this effect.
Does anyone know why ?
(by the way : I'm using the pov31b6 msdos version)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
H.A. van der Meiden wrote:
>
> Dan Connelly wrote:
> > blah, blah, blah....
> You're right !
Miracles do happen....
> After some trial it seems that the absorption parameter is an
> exponentional falloff rate.
> ...
> Unfortunatly high values for both absorption and emission (>2.0) result
> in noisy images, although high sample rates seem reduce this effect.
> Does anyone know why ?
I saw the same thing. Think about it.... the code just randomly
sample the material using some magical algorithm. The order in which
it does absorption and emission events has a big effect if each
event has a large impact.... the one done closest to the
surface wins. Thus
(material depth / sampling count) min[absorption number , emission number]
should be much less than one if the effect is to be avoided.
But then, I've already more than used up my "correct statement" quota
today :).
Dan
--
http://www.flash.net/~djconnel/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
H.A. van der Meiden wrote in message <360### [at] twi tudelftnl>...
>Whilst expirimenting with the media feature, I've noticed something odd.
[snips]
> interior
> {
> media
> {
> samples 3,10
> absorption rgb 1
> emission rgb <0,0,1>
> scattering {0, rgb <0,0,5>
> extinction 0.0}
> }
> }
This scattering call will generate a non-fatal warning in Pov3.1b6
Easy enough to overlook that warning ....
Re-evaluate your results after putting in an actual scattering type.
des
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
