Hello,

i have been wondering whether the recent 3.5 version of POVray
features a different attenuation formula in comparison to 3.1a
(that's the version i have) ...
Well, actually, the formula in 3.1a is not bad, as it approximates
reality quite well, but it is not 100% correct ... it should be an
exponential decay - something like Inew = Iold * (e ** (- k * d)),
with e = 2.718.., ** = power, k = constant determining the strength
of attenuation, d = distance the ray had to take through the medium -
instead, shouldn't it?


Regards,

Post a reply to this message

"Niels Boehm" <boh### [at] minetuni-jenade> wrote in message
news:Pin### [at] paxp12mipooluni-jenade...

i have been wondering whether the recent 3.5 version of POVray
features a different attenuation formula in comparison to 3.1a
(that's the version i have) ...

It has carried over the "more realistic attenuation" from MegaPOV 0.7.
Use it by specifying a number of 1000 or greater for 'fade_power'.

Well, actually, the formula in 3.1a is not bad, as it approximates
reality quite well, but it is not 100% correct ... it should be an
exponential decay - something like Inew = Iold * (e ** (- k * d)),
with e = 2.718.., ** = power, k = constant determining the strength
of attenuation, d = distance the ray had to take through the medium -
instead, shouldn't it?

Don't ask me.  ;-)
The doc says:  Attenuation = exp(-depth/fade_dist)
However, that's only for 'interior' and not for lights.  Light sources still
only use the usual linear, quadratic, etc., fading.

Bob H.

Post a reply to this message

"Bob H." wrote:
> Light sources still
> only use the usual linear, quadratic, etc., fading.

Of which the fade_power 2 is the physically accurate one. Actually
it is calculated so that it is accurate for a light source with
non zero dimensions which is better than 1/r^2 because most real
light sources are not point sources.

I'm anticipating some responses to this so here is some more. Light
_does_not_ fade in vacuum because there is some mysterious "ether"
in there. The fading is caused by simple geometry. Let's say a point
source at the center of a 1 m radius sphere emits 1 W of energy. The
inside of the sphere receives 1 W of energy. The surface area is

1 W of energy (no ether). The surface brightness of the sphere has
decreased because the energy per area has decreased. The area is now

This where the 1/r^2 comes.

Even lighting engineers don't bother to calculate the effect of
absobtion caused by air but it's possible in povray. Put
media_attenuation on in the light sources and some media in the scene.
The effect is visible over large distances, like 1 km.


_____________
Kari Kivisalo

Post a reply to this message

On Sat, 8 Sep 2001, Bob H. wrote:

> "Niels Boehm" <boh### [at] minetuni-jenade> wrote in message
> news:Pin### [at] paxp12mipooluni-jenade...
> 
> > i have been wondering whether the recent 3.5 version of POVray
> > features a different attenuation formula in comparison to 3.1a
> > (that's the version i have) ...
> 
> It has carried over the "more realistic attenuation" from MegaPOV 0.7.
> Use it by specifying a number of 1000 or greater for 'fade_power'.

Well, then I'll have to wait until 3.5 possibly gets ported to my 
platform ;)

> > Well, actually, the formula in 3.1a is not bad, as it
> > approximates reality quite well, but it is not 100% correct ...
> > it should be an exponential decay - something like Inew = Iold *
> > (e ** (- k * d)), with e = 2.718.., ** = power, k = constant
> > determining the strength of attenuation, d = distance the ray had
> > to take through the medium - instead, shouldn't it?
> 
> Don't ask me.  ;-)
> The doc says:  Attenuation = exp(-depth/fade_dist)

Yeah, that's virtually the same formula as i suggested ;)
(with k = 1/fade_dist)

> However, that's only for 'interior' and not for lights.  Light sources still
> only use the usual linear, quadratic, etc., fading.

Well, the attenuation in the interior of objects was exactly what i
was interested in, since the other stuff is quite correct as it is.


Thanks,
Niels Boehm

Post a reply to this message