I'm experimenting with clouds. The attached image demonstrates. (The
view under the clouds is looking north along the western shore of the
Sea of Smyth, on the eastern edge of the moon --- the big crater is
probably Napier. The camera is 200km up.)

The clouds are implemented as a textured sphere 35km above the surface.
The cirrus texture on the sphere is adequate, but I can fix that. The
thing I'm worried about is the dark edge on the horizon.

I know what this is: my sphere is defined as:

	sphere
	{
		<0, 0, 0>, 1
		hollow
		double_illuminate

		texture {
			pigment {
				color_map {
					[0 rgbt <1, 1, 1, 1>]
					[1 rgbt <1, 1, 1, 0>]
				}
			}

			finish {
				diffuse 1
			}
		}

		scale r
	}

Where the color_map is selecting a partially transparent color, then
we're always going to be blocking some of the light coming through,
therefore it's going to be darker.

What's a good way round this? Should I be using reflection for this,
rather than just a simple pigment colour? That would give me access to
the huge zoo of reflection options, including the energy conservation
feature. Is there an easier way?

-- 
┌─── ｄｇ＠ｃｏｗｌａｒｋ．ｃｏｍ ─────
http://www.cowlark.com ─────
│ "There does not now, nor will there ever, exist a programming
│ language in which it is the least bit hard to write bad programs." ---
│ Flon's Axiom

Post a reply to this message

Attachments:
Download 'clouds.jpg' (60 KB)

Preview of image 'clouds.jpg'

Am 18.12.2013 01:27, schrieb David Given:
> I'm experimenting with clouds. The attached image demonstrates. (The
> view under the clouds is looking north along the western shore of the
> Sea of Smyth, on the eastern edge of the moon --- the big crater is
> probably Napier. The camera is 200km up.)
>
> The clouds are implemented as a textured sphere 35km above the surface.
> The cirrus texture on the sphere is adequate, but I can fix that. The
> thing I'm worried about is the dark edge on the horizon.
>
> I know what this is: my sphere is defined as:

No, I'm pretty sure that isn't the reason; this must be something different.

I think what's really happening is that your sphere cuts away the media. 
I'm not sure how media is handled in such cases, but my guess is that 
the inner sphere "overrides" the media, and that the "hollow" keyword 
only serves to make the sphere available as another media container.

Try explicitly adding the atmosphere media to the cloud sphere.

Post a reply to this message

Am 18.12.2013 02:17, schrieb clipka:
> Am 18.12.2013 01:27, schrieb David Given:
>> I'm experimenting with clouds. The attached image demonstrates. (The
>> view under the clouds is looking north along the western shore of the
>> Sea of Smyth, on the eastern edge of the moon --- the big crater is
>> probably Napier. The camera is 200km up.)
>>
>> The clouds are implemented as a textured sphere 35km above the surface.
>> The cirrus texture on the sphere is adequate, but I can fix that. The
>> thing I'm worried about is the dark edge on the horizon.
>>
>> I know what this is: my sphere is defined as:
>
> No, I'm pretty sure that isn't the reason; this must be something
> different.

Tests with a simple sample scene show that I was wrong as well: The 
"hollow" keyword does essentially work as you expected. Instead, the 
problem is somehow related to the way the media is sampled. Severely 
cranking up the number of samples reduces the effect to invisibility.

Post a reply to this message

On 18/12/13 01:53, clipka wrote:
[...]
> Tests with a simple sample scene show that I was wrong as well: The
> "hollow" keyword does essentially work as you expected. Instead, the
> problem is somehow related to the way the media is sampled. Severely
> cranking up the number of samples reduces the effect to invisibility.

I'm very sorry to hear that --- my renders already take an age. I've
tried the atmospheric media with a samples setting of 60 and the
artifact is still present.

It does occur to me that I'm going to need another media anyway, for
when I add cumulus clouds low down. So I might try to wrap the cirrus
clouds into that. If I make the clouds media the same size as the
atmospheric media then I should avoid any embarrassing edges. It's a bit
of a waste given that the atmosphere is 200km thick, but it'll probably
work.

-- 
┌─── ｄｇ＠ｃｏｗｌａｒｋ．ｃｏｍ ─────
http://www.cowlark.com ─────
│ "There does not now, nor will there ever, exist a programming
│ language in which it is the least bit hard to write bad programs." ---
│ Flon's Axiom

Post a reply to this message

On 18/12/13 15:19, David Given wrote:
[...]
> It does occur to me that I'm going to need another media anyway, for
> when I add cumulus clouds low down. So I might try to wrap the cirrus
> clouds into that. If I make the clouds media the same size as the
> atmospheric media then I should avoid any embarrassing edges. It's a bit
> of a waste given that the atmosphere is 200km thick, but it'll probably
> work.

No, that doesn't work. The atmospheric shell is 200km thick, but the
cloud layer within it is only 1km thick. This means that unless I'm
looking at an oblique angle through the clouds, then chances are that no
samples will hit the clouds, and so Povray thinks the media is empty.
The end result is that I can see clouds a long way off but the camera's
surrounded by a circle of blue sky.

I can crank up the number of samples but I still end up with weird
artifacts (and very slow renders). This is all being implemented as a
single object with two medias: oddly, the number of samples for the
first media is being used for both. The samples setting for the second
media is ignored completely...

Picture enclosed. It's interesting artistically, but that's supposed be
a completely homogeneous shell 1km thick...

-- 
┌─── ｄｇ＠ｃｏｗｌａｒｋ．ｃｏｍ ─────
http://www.cowlark.com ─────
│ "There does not now, nor will there ever, exist a programming
│ language in which it is the least bit hard to write bad programs." ---
│ Flon's Axiom

Post a reply to this message

Attachments:
Download 'newmoon.jpg' (84 KB)

Preview of image 'newmoon.jpg'

Am 18.12.2013 19:37, schrieb David Given:

> I can crank up the number of samples but I still end up with weird
> artifacts (and very slow renders). This is all being implemented as a
> single object with two medias: oddly, the number of samples for the
> first media is being used for both. The samples setting for the second
> media is ignored completely...

If the first media is the one with the higher settings, that's intentional.

Post a reply to this message

Am 18.12.2013 01:27, schrieb David Given:

> The clouds are implemented as a textured sphere 35km above the surface.
> The cirrus texture on the sphere is adequate, but I can fix that. The
> thing I'm worried about is the dark edge on the horizon.

Should you happen to be running Linux, you might want to try the newest 
code from GitHub:

https://github.com/POV-Ray/povray

Post a reply to this message

Am 20.12.2013 13:54, schrieb clipka:
> Am 18.12.2013 01:27, schrieb David Given:
>
>> The clouds are implemented as a textured sphere 35km above the surface.
>> The cirrus texture on the sphere is adequate, but I can fix that. The
>> thing I'm worried about is the dark edge on the horizon.
>
> Should you happen to be running Linux, you might want to try the newest
> code from GitHub:
>
> https://github.com/POV-Ray/povray

... and if you happen to be running Windows, one of these (unofficial!) 
binary builds might get you going:

https://github.com/c-lipka/povray/releases/tag/v3.7.0.20131220

The files are drop-in replacements for pvengine32.exe, 
pvengine32-sse2.exe and pvengine64.exe, respectively.

Post a reply to this message

On Wed, 18 Dec 2013 00:27:37 +0000
David Given <dg### [at] cowlarkcom> wrote:

> The clouds are implemented as a textured sphere 35km above the surface.
> The cirrus texture on the sphere is adequate, but I can fix that. The
> thing I'm worried about is the dark edge on the horizon.

Just an observation - clouds are generally from 0 to say 10 or 12 km.
Normally the height of the clouds is only 4-6km. Only in summertime,
you'l have cumulonimbus reaching above 10km. Clouds at 35 km do exist,
but are normally not directly visible.

Also, air density dimishes exponentially. It's 1000 hpa at 0 meters,
and only 6 hpa at 30 km. So you'll have to simulate the varying 
refraction index. http://jcoppens.com/globo/teoria/atmcalc.en.php
has still a calculator from the time we launched a few experimental
balloons.

This might help your realism a little.

Post a reply to this message

Hi(gh)!

On 24.12.2013 05:47, John Coppens wrote:
> On Wed, 18 Dec 2013 00:27:37 +0000
> David Given <dg### [at] cowlarkcom> wrote:
>
>> The clouds are implemented as a textured sphere 35km above the surface.
>> The cirrus texture on the sphere is adequate, but I can fix that. The
>> thing I'm worried about is the dark edge on the horizon.
>
> Just an observation - clouds are generally from 0 to say 10 or 12 km.
> Normally the height of the clouds is only 4-6km. Only in summertime,
> you'l have cumulonimbus reaching above 10km.

You have to take into account that the scene is set on a terraformed 
Earth's moon - with only one sixth of Earth's gravity, the atmosphere is 
much less compressed towards the surface than on Earth (for example: on 
Saturn's moon Titan, where gravity is slightly lower than on Earth's 
moon, the atmosphere reaches as high as almost 1,000 kms above the 
surface, and clouds hover at altitudes between 15 and 40 kms), so that 
the tropopause (upper limit for clouds) will be much higher than on Earth.

 > Clouds at 35 km do exist,
 > but are normally not directly visible.

Did you think of polar stratospheric clouds or of noctilucent clouds?


See you in Khyberspace!

Yadgar

Now playing: The Turn Of A Dream (Robert Schroeder)

Post a reply to this message