"Chris R" <car### [at] comcastnet> wrote:

XD Eyeball-science and odometrics ! And in the end it always ends up in sensory
captation before the brain. we are the weak chain link.


> So, to make my macros look more technically astute, I have just added a "Dark
> Adaptation" color value when computing the color of the moon's light before
> applying the lux value and atmospheric attenuation.  It still means I'm mostly
> just adjusting things until it looks right, but I can pretend there's some math
> and science behind it.  :-)
>
> -- Chris R.

Post a reply to this message

Le 2022-03-23 à 16:06, Chris R a écrit :
> Cousin Ricky <ric### [at] yahoocom> wrote:
>> On 2022-03-21 10:58 (-4), Chris R wrote:
>>> I am working on a scene that is lit by moonlight coming through a window with
>>> partially open blinds.  I use LightSys, and have set up the light source as an
>>> area light just outside of the window itself.
>>>
>>> I was wondering if anyone had suggestions for good color/lumens combinations to
>>> simulate moonlight.  I am currently using a made-up color of <0.45, 0.5, 0.75>
>>> and an intensity of 1.0 with no fading, (so I guess I'm not really using
>>> LightSys to great effect here).  Has anyone done any spectrum analysis for
>>> moonlight to suggest better values?
>>
>> Lumens is not what you want for moonlight; this is a measure of the
>> total amount of light emitted (or reflected, in this case).  Lux is the
>> proper unit for a light source that is effectively at infinity.
>>
>> Wikipedia says 0.05 to 0.3 lux.  That's quite a range!
>>
>> To get a narrower precision, a preliminary calculation is:
>>    solar lux * lunar albedo * (solar radius / astronomical unit) ^ 2
>>
>> Given 98,000 lux for overhead sunlight and an albedo of 7% for the Moon,
>> this yields 0.148 lux.
>>
>> This should be multiplied by the ratio of the apparent areas of the Moon
>> and the Sun, but since they are approximately the same size in our sky,
>> that figure cancels.  The number should be further reduced because the
>> full Moon is not a flat reflector.  If I'm thinking correctly (not a
>> safe bet), a Lambertian reflector would halve the light; since the Moon
>> is not a Lambertian reflector, the true lux is somewhere between 0.074
>> and 0.148.
>>
>> Finally, the closer the Moon to the horizon, the more the atmospheric
>> attenuation.  My previous calculations for sunlight (and moonlight
>> should be similar), assuming low turbidity, were:
>>    96% at 60 degrees
>>    86% at 45 degrees
>>    77% at 30 degrees
>>    47% at 15 degrees
>>    0% at moonrise
>> Obviously, those figures close to the horizon are too small, and that is
>> because I didn't consider the curvature of the Earth's atmosphere. My
>> guess, without consulting my musty calculus textbooks, is that the
>> attenuation is about 7-8% at moonrise.
>>
>> Of course, with the Moon so far away, you do /not/ want fading; and you
>> should use a parallel light.
>>
>> I have questions about the spectral data given for the moon rocks in
>> Lightsys IV, as the Moon does not appear blue to me; and this, even with
>> the blue bias that my vision appears to have in low light.  I have not
>> searched for better spectral data, but even with such data, you would
>> need to factor in solar spectral data (because the Moon only reflects
>> light) and atmospheric attenuation (ouch!).  Best I have done is to
>> compare the Moon to streetlights and indoor lighting, and I have
>> concluded that moonlight has a color temperature of around 4000 K when
>> the Moon is moderately high in the sky.  Lightsys IV macro Blackbody()
>> can calculate that color for you; remember to decide what white balance
>> you want.
> 
> Going back to this again, I realized there were a number of errors I introduced
> when I moved from my guess at moon light and calculations using Lux and
> including a parallel, circular area light.  So, I'd like to go back to your
> calculations above, which give you a Lux value for the moon in various
> configurations, and determine how to combine that with the Blackbody() macro for
> the moon's light temperature?  In LightSys, the light color gets multiplied by
> the lumens, and then light fading takes care of reducing that back to normal
> brightness levels.  When using LightSys for the sun, as a non-fading light,
> small brightness constants are used instead of lumens, and I pick them
> arbitrarily, (1.0, 2.0, ...).
> 
> Should I just be multiplying the Blackbody() color by the calculated lux value?
> I've tried that, and it seems to be too dim based on the light I have seen
> coming through the skylights in my house on a full moon, so I am wondering if I
> am missing something.  For now, I just introduce a brightness fudge factor,
> similar to what I do with the sun, but that doesn't make me happy.  Of course, I
> could also be misperceiving the brightness of the real moon coming through my
> windows for other reasons, too, and the dimmer color is more accurate.
> 
> Any thoughts?
> 
> -- Chris R.
> 
> 

Our eyes will adapt to the much lower illumination of the Moon, making 
it look much brighter than it actually is.
Your setup show the actual illumination. You want to show the perceptual 
illumination. So, you need to increase your illumination accordingly. 
You may need to make your Moon up to 10 to 50 times brighter.

Post a reply to this message

"Chris R" <car### [at] comcastnet> wrote:
> "Kenneth" <kdw### [at] gmailcom> wrote:
> > Cousin Ricky <ric### [at] yahoocom> wrote:
> > > On 2022-03-21 10:58 (-4), Chris R wrote:
> > > >
> > > > ...I am currently using a made-up color of <0.45, 0.5, 0.75>
> > > > and an intensity of 1.0 with no fading...Has anyone done any spectrum
> > > > analysis for moonlight to suggest better values?
> > >
> > > Lumens is not what you want for moonlight; this is a measure of the
> > > total amount of light emitted (or reflected, in this case).  Lux is the
> > > proper unit for a light source that is effectively at infinity...
> > > [clip]
> >
> > That's a masterful analysis; thanks for taking the time to describe the details
> > so clearly.
> >
> > I find it interesting that the full Moon as seen at zenith (with just our eyes,
> > not through a telescope) *appears* to be a Lambertian 'disc', not a sphere--
> > that is, it seems to have equal illuminance from all points on its surface...
> > even though the Sun's rays are hitting the sphere at progressively shallower
> > angles, out towards the rim. I'm guessing that this is a result of our automatic
> >  eye/brain adjustments having to do with 'local contrast' against a dark sky,
> > and because of the behavior of our eyes' receptors in *relatively* dim light.
> >
> > The color temperature of Moonlight, as reflected from objects, seem to be
> > somewhat subjective from what I've read so far. We think of it as being slightly
> > blue-ish-- but the Moon itself, reflecting Sunlight, has a LOWER color
> > temperature than the Sun, something like 4100-deg Kelvin. Taken by itself, that
> > would mean that Moonlight should be slightly *yellowish*(!) in comparison. But
> > other things come into play-- the Earth's atmospheric scattering, our eyes'
> > reaction to dim light, and probably other mysteries that I don't know about...
> > all of which combine to create a slightly blue-ish result.  It's also possible
> > that we have been 'trained' to think of Moonlight as being blue-- from centuries
> > of artists' interpretations, how it is usually reproduced in movies, etc. No one
> > wants to see yellow-ish Moonlight, it just looks... wrong! ;-)
> >
> > Take a look at this discussion (although I don't think it comes to any definite
> > conclusion)...
> >
> >
https://physics.stackexchange.com/questions/244922/why-does-moonlight-have-a-lower-color-temperature
> >
> > Here's another little monkey-wrench to throw into the mix: the D65 'white point'
> > of computer monitors and some color standards. That's a 6500-deg Kelvin color
> > temperature, which is meant to match 'daylight'-- meaning, the Sun's light AND
> > the surrounding blue sky. That's more blue-ish than the Sun by itself (which is
> > around 5400 to 5700K?) But does this viewing 'environment' still hold true when
> > trying to create a 'low-light' Moonlit scene in POV-ray (or any other graphics
> > program)? By nature, such a scene should be reproduced very dim and dark to look
> > 'natural'-- compared to a 'Sun-lit' scene. But that's not how we would usually
> > render it, because we want it to look...normal. Meaning, colors and brightnesses
> > rendered from 0 to 1.0 (0 to 255) or thereabouts-- just like a typical scene in
> > 'daylight'. This...discrepancy(?) in how we render low-light scenes and their
> > associated 'color temperatures' seems like an interesting conundrum. Do I have a
> > solution to it? No! (ha). I just thought I would bring it up as a philosophical
> > discussion point. Maybe we should just follow the great artists of the past, and
> > color the Moonlight blue regardless!
>
> So, I'm feeling less bad about my arbitrary adjustments to the moon's light
> color now.  I found a whole slew of information about the eye's dark adaptation
> ability, and what it can do to our perception of color and brightness,
> (specifically the Purkinje effect causing a shift towards blue sensitivity,
> which provides another explanation for our perception of moonlight as blue.)
>
> So, to make my macros look more technically astute, I have just added a "Dark
> Adaptation" color value when computing the color of the moon's light before
> applying the lux value and atmospheric attenuation.  It still means I'm mostly
> just adjusting things until it looks right, but I can pretend there's some math
> and science behind it.  :-)
>
> -- Chris R.

I just posted some renderings of the scene with the moonlight as light sources
in p.b.i:

http://news.povray.org/povray.binaries.images/thread/%3Cweb.623cb1d52227bb25cc71334a5cc1b6e@news.povray.org%3E//povray.
binaries.images/thread/%3Cweb.623cb1d52227bb25cc71334a5cc1b6e@news.povray.org%3E/


-- Chris R.

Post a reply to this message

On 2022-03-24 (-4) 00:27, Cousin Ricky wrote:
> 
> Alas, I do not have a simple explanation for how this works; I don't
> completely understand it myself.  I just examined test_lightsys.pov, and
> used the technique on a solar spectrum to get my light_source colors.
> Lightsys IV can do white point adjustments without requiring a spectal
> curve, but I haven't figured out how to do that yet.  Have a look at
> scenes im_test_day.pov and im_test_inc.pov to see if that helps.

Oops, I mistook the image map filenames for the scene filename.  The
scene description file is demo_image_map.pov, in the demos subdirectory
of Lightsys IV.

Post a reply to this message