On 03/13/2018 07:39 AM, Norbert Kern wrote:
> William F Pokorny <ano### [at] anonymousorg> wrote:
> 
> Hi William,
> 
> scene took about 5 days to render with 8000*4500 pixels.
> This worked because I divided the job in three parts - the clouds, a render with
> low radiosity settings (clouds as image_map) and a part render of the head with
> high radiosity settitngs.
> 
> The clouds are df3 media ones by Gilles Tran - see
> http://www.oyonale.com/modeles.php?lang=en&page=36 . He also presented there a
> brief summary of the available methods to create clouds in POV-Ray.
> My only contribution are extreme coloring and a high Intervals value (7 instead
> of 2-5).
> 
> I put a reduced scene file to pbs-f.
> 
> Norbert
> 

Thank you for posting your code I can review and with which I can 
experiment.

I'd played some with Gilles' cloud scene some years back and used a 
variation of his df3 technique to create a Halloween ghost using only 
emissive media. I'm somewhat familiar with the method.

And speaking of emissive media, I see there is:

  emission C_Emission

in your media {} code that I don't recall seeing in the original cloud 
generator method. C_Emission being set to a sky color but at very low 
intensity. Is this something you added?

I've yet to look in detail at media using scattering, absorption and 
emission. The radiosity block has media off so suppose the emissive 
component affecting only the cloud resultant color's in some fashion. 
I'll try to get in a couple renders with and without the emission color 
later today to see what differences pop out.

There is too fog{} I see.

Bill P.

Post a reply to this message

William F Pokorny <ano### [at] anonymousorg> wrote:
>
> Thank you for posting your code I can review and with which I can
> experiment.
>
> I'd played some with Gilles' cloud scene some years back and used a
> variation of his df3 technique to create a Halloween ghost using only
> emissive media. I'm somewhat familiar with the method.
>
> And speaking of emissive media, I see there is:
>
>   emission C_Emission
>
> in your media {} code that I don't recall seeing in the original cloud
> generator method. C_Emission being set to a sky color but at very low
> intensity. Is this something you added?
>
> I've yet to look in detail at media using scattering, absorption and
> emission. The radiosity block has media off so suppose the emissive
> component affecting only the cloud resultant color's in some fashion.
> I'll try to get in a couple renders with and without the emission color
> later today to see what differences pop out.
>
> There is too fog{} I see.
>
> Bill P.


"emission C_Emission" is only used in "makecloud2_demo_sunset.pov" and
"makecloud2_demo_sunset2.pov". The latter was used as a starting point for my
scene.
It didn't catch my eyes.

I remember your image - very impressive.
I've to make something in that direction too...

Norbert

Post a reply to this message

> William F Pokorny <ano### [at] anonymousorg> wrote:
>>
>> Impressive! Especially the sky & clouds.
>>
>> POV-Ray media I assume? Fog for the horizon or was that done as part of
>> the overall media?
>>
>> If yes to the first question, could you offer a brief description of the
>> set up? Multiple containers, one container or global media? How was the
>> density defined? Scattering/absorption or emulated with emissive?
>>
>> I've been wandering around in the media code of late so I'm interested...
>>
>> Bill P.
> 
> Hi William,
> 
> scene took about 5 days to render with 8000*4500 pixels.
> This worked because I divided the job in three parts - the clouds, a render with
> low radiosity settings (clouds as image_map) and a part render of the head with
> high radiosity settitngs.
> 
> The clouds are df3 media ones by Gilles Tran - see
> http://www.oyonale.com/modeles.php?lang=en&page=36 . He also presented there a
> brief summary of the available methods to create clouds in POV-Ray.
> My only contribution are extreme coloring and a high Intervals value (7 instead
> of 2-5).
> 
> I put a reduced scene file to pbs-f.
> 
> Norbert
> 
> 
> 

Using intervals 1 and tripling the samples value would have been much 
faster.
You ONLY use intervals >1 when using method 1 or 2, never ever with 
method 3 (default).

It probably have taken less than a day using intervals 1 and 10 times as 
many samples.

Post a reply to this message

> Hi William,
> 
> scene took about 5 days to render with 8000*4500 pixels.
> This worked because I divided the job in three parts - the clouds, a render with
> low radiosity settings (clouds as image_map) and a part render of the head with
> high radiosity settitngs.
> 
> The clouds are df3 media ones by Gilles Tran - see
> http://www.oyonale.com/modeles.php?lang=en&page=36 . He also presented there a
> brief summary of the available methods to create clouds in POV-Ray.
> My only contribution are extreme coloring and a high Intervals value (7 instead
> of 2-5).
> 
> I put a reduced scene file to pbs-f.
> 
> Norbert
> 
> 
> 

Taking a better look at your clouds' media:
Using intervals 7 is absolutely killer for the render time.
Default is : intervals 1
You should never change that value!

You use the default samples value of 10, which is OK as a starting 
point. Increase if needed.

If you need more precision, you should always increase the samples 
value, never the intervals value.

Don't forget that, if you use 2 values for the samples count, the second 
value is always silently ignored.

In this case, using :
intervals 1// default value
samples 100

Your media would have been about 10 times faster to render.
So, I'd expect a render time between 12 and 24 hours, and have better 
quality for the media.

In your Tree() macro, you only get a single sample because you use 
samples 1,1. The default is samples 10.
That sample is taken at, or near, the mid point through the container.

Post a reply to this message

On 3/11/2018 2:41 PM, Norbert Kern wrote:
> a comment on the present conditio humana.
> 
> 
> Norbert
> 


Wow, very nice! I like the poofy-yet-hazy clouds and the grass especially.

:)


Mike

Post a reply to this message

On 03/13/2018 09:48 AM, Norbert Kern wrote:
> William F Pokorny <ano### [at] anonymousorg> wrote:
> 
> "emission C_Emission" is only used in "makecloud2_demo_sunset.pov" and
> "makecloud2_demo_sunset2.pov". The latter was used as a starting point for my
> scene.
> It didn't catch my eyes.

Ah, I see now too Gilles did have it there. Thanks.

I got a couple small renders done yesterday - with and without the 
emission - and compared them. At a high level adding in emission the 
color of the background sky makes the clouds look as if they are sitting 
in a hazy atmosphere without having modeled the hazy atmosphere. See the 
attached image.

The difference image on the right shows an aspect or pitfall of the 
method - depending on one's goal I suppose. Once one or more color 
channels for the media 'interval' being sampled is maxed out (Red here), 
you get a 'disjoint' color-hue change in the media due the emission 
adder.

> 
> I remember your image - very impressive.

Thanks - though I certainly don't play in your artistic league. :-)

Bill P.

Post a reply to this message

Attachments:
Download 'e_to_noe.jpg' (43 KB)

Preview of image 'e_to_noe.jpg'

"Norbert Kern" <nor### [at] t-onlinede> wrote:
> a comment on the present conditio humana.
>
>
> Norbert

Wow, this is an amazing scene. Making me want to dust off my povray skills (hope
I haven't forgotten too much).

Sean

Post a reply to this message

Norbert Kern wrote on 13/03/2018 12:53:
> Paolo Gibellini <p.g### [at] gmailcom> wrote:
> 
>> A significant scene.
>> Paolo
> 
> 
> 
> Many thanks, Paolo.
> 
> When will we see some new images from you?
> 
> Norbert
> 
> 
> 
Ha! You are right...
I've changed somehow my job and the pov-time has been absorbed into the 
new tasks, but I have not stopped putting aside new ideas.
Before the end of the spring I should be able to post something.
;-)
Paolo

Post a reply to this message

Alain <kua### [at] videotronca> wrote:
>
> Using intervals 1 and tripling the samples value would have been much
> faster.
> You ONLY use intervals >1 when using method 1 or 2, never ever with
> method 3 (default).
>
> It probably have taken less than a day using intervals 1 and 10 times as
> many samples.


You are right with the intervals 1 issue.
If it wasn't a Gilles Tran code, i'ld have checked more carefully...

Meanwhile I rendered with intervals 1 and samples 70.
It's not faster, but the result is slightly better.

Norbert

Post a reply to this message

"s.day" <s.d### [at] uelacuk> wrote:

> Wow, this is an amazing scene. Making me want to dust off my povray skills (hope
> I haven't forgotten too much).
>
> Sean


Hi Sean,

it's time to come back...

Norbert

Post a reply to this message