"Thomas de Groot" <t.d### [at] internlnet> wrote:
> Nice work, Bill!
Thanks!

> You are on the right track. Still a bit *cotton*-like I believe, not
> entirely cloud-like. How is the overlapping container issue behaving?
> Gilles' macro suffers from that unfortunately.
Yes, this is an issue since I am following Gilles' recipe almost to the
letter... and one must also remember to increase max_trace_level!

"St." <dot### [at] dotcom> wrote:
> That's very nice work Bill, but to me, it has a 'painted' look to it at the
> moment. And if I say painted, I mean in almost professional terms.
I think that's the low quality media settings talking. I'll try with more
intervals and samples and see how it compares.

"Joanne Simpson" <cor### [at] onewhiteravencom> wrote:
> Hi Bill,
> could you post your media settings? I might have a scene to go with your
> clouds ;-)

Certainly! Check out Gilles' cloud examples too - I think his look much
better than mine.

#declare Cloud1 = box {
  -0.5, 0.5
  pigment { color Clear }
  finish { ambient 0 diffuse 0 }
  hollow
  interior {
    media {
      scattering { 1, 0.0275 extinction 1 }
      absorption 0.02
      emission 0.01
      intervals 3
      samples 10,100
      density {
        density_file df3 "CloudTall.df3"
        interpolate 1
        translate -0.5
        scale <1, -1, 1> } } }
  scale 150*<4/3, 1, 4/3> }

This is a higher quality version. Set intervals to 1 and comment out samples
for the version I posted above. If you want the whole scene let me know;
you'll also need some suitable dxf files. They're about 9MB each, so I
can't really post them!

Bill

Post a reply to this message

And another with one layer (not enough samples ..) and 2 individual clouds.

Post a reply to this message

Attachments:
Download 'sample_27.jpg' (31 KB)

Preview of image 'sample_27.jpg'

"Bruno Cabasson" <bru### [at] alcatelaleniaspacefr> wrote:
> I recently worked on the question of clouds
Hi Bruno!

I have to say, these are awesome clouds. If I can get my cumulus looking
half as good as this... :) Most of these images appear to have clouds as
continuous layers - are these isosurface clouds? And I notice there are
some 'fringing' glows when the sun is behind the clouds - does this
behaviour occur naturally when you move the light source or does it require
different media?

> and I found that one should
> separate cloud layers from individual clouds.
I'm not sure I understand what you mean by this... do you mean to use either
one or the other?

> But I would not recomment adjusting the
> aspect via emission. First because it does not correspod to reality, and
> second because it will interact with radiosity in a wrong manner.
Yes, I'd rather not have used emission, but it did make a big difference. As
to reality - well, the effect I'm looking for would have to be simulated by
media interacting with itself via radiosity, which I don't believe is
possible.

> Btw: can I attach more than one image in a post?
Not via the web interface. With a newsreader, I believe so.

> Now I am working on some other features: sun aspect (halo ...)
I'd like to see that when you have some results!

Bill

Post a reply to this message

"Bill Pragnell" <bil### [at] hotmailcom> wrote:
> "Bruno Cabasson" <bru### [at] alcatelaleniaspacefr> wrote:
> > I recently worked on the question of clouds
> Hi Bruno!
>
> I have to say, these are awesome clouds. If I can get my cumulus looking
> half as good as this... :) Most of these images appear to have clouds as
> continuous layers - are these isosurface clouds? And I notice there are
> some 'fringing' glows when the sun is behind the clouds - does this
> behaviour occur naturally when you move the light source or does it require
> different media?
>

No they are scattering media between two planes (for one layer). I use
built-in patterns (bozo, crackle, wrinkles, spiral, ...) Scattering type is
2 (MIE_HAZY_SCATTERING), which corresponds to natural clouds and to what is
said in the documentation about scattering (3.6.2.1.3  Scattering). This
produces the effect whith the sun. But I can chose the scattering tpe

Until now, I can define 3 independent layers, and 5 individual clouds, plus
a specific layer for some kind of oveall mist (plus fog of course, plus
another specific dispersive layer ...). It seems to be enough for most
cases. But I am still in trial phase, improving and trying to find an
easy-to-use way to make acceptable clouscapes. I got inspired by Terragen.

One of the problems was to contol the turbulence parameters. And my model is
not really suitable for large fat and round cumuli that have a flat base,
round turbulences, and can go high in the verical dimension. Though my
individual clouds do have a flat base and might look like cumuli. But
observing nature makes me see I and lightyears from it.

> > and I found that one should
> > separate cloud layers from individual clouds.
> I'm not sure I understand what you mean by this... do you mean to use either
> one or the other?

I ment that I never could make good cloudscapes with individual clouds. And
I frequently encounter problems with the containers when they overlap,
mostly if the whole scene is also surrounded by media.
>
> > But I would not recomment adjusting the
> > aspect via emission. First because it does not correspod to reality, and
> > second because it will interact with radiosity in a wrong manner.
> Yes, I'd rather not have used emission, but it did make a big difference. As
> to reality - well, the effect I'm looking for would have to be simulated by
> media interacting with itself via radiosity, which I don't believe is
> possible.
>
> > Btw: can I attach more than one image in a post?
> Not via the web interface. With a newsreader, I believe so.
>
> > Now I am working on some other features: sun aspect (halo ...)
> I'd like to see that when you have some results!
>
> Bill

Post a reply to this message

Your clouds are VERY impressive!!

Thomas

Post a reply to this message

"Thomas de Groot" <t.d### [at] internlnet> wrote:
> Your clouds are VERY impressive!!
>
> Thomas

Tanks.
Here is this afteroon's render: 2h30mn. One high layer (10000m high, 150m
thick, used agate pattern), and an individual cloud (intended to be a
cumulus).

Regards

Post a reply to this message

Attachments:
Download 'sample_30.jpg' (25 KB)

Preview of image 'sample_30.jpg'

Bill Pragnell nous apporta ses lumieres en ce 25/08/2006 06:12:
> "Thomas de Groot" <t.d### [at] internlnet> wrote:
>> Nice work, Bill!
> Thanks!
> 
>> You are on the right track. Still a bit *cotton*-like I believe, not
>> entirely cloud-like. How is the overlapping container issue behaving?
>> Gilles' macro suffers from that unfortunately.
> Yes, this is an issue since I am following Gilles' recipe almost to the
> letter... and one must also remember to increase max_trace_level!
> 
> "St." <dot### [at] dotcom> wrote:
>> That's very nice work Bill, but to me, it has a 'painted' look to it at the
>> moment. And if I say painted, I mean in almost professional terms.
> I think that's the low quality media settings talking. I'll try with more
> intervals and samples and see how it compares.
> 
> "Joanne Simpson" <cor### [at] onewhiteravencom> wrote:
>> Hi Bill,
>> could you post your media settings? I might have a scene to go with your
>> clouds ;-)
> 
> Certainly! Check out Gilles' cloud examples too - I think his look much
> better than mine.
> 
> #declare Cloud1 = box {
>   -0.5, 0.5
>   pigment { color Clear }
>   finish { ambient 0 diffuse 0 }
>   hollow
>   interior {
>     media {
>       scattering { 1, 0.0275 extinction 1 }
>       absorption 0.02
>       emission 0.01
>       intervals 3
>       samples 10,100
>       density {
>         density_file df3 "CloudTall.df3"
>         interpolate 1
>         translate -0.5
>         scale <1, -1, 1> } } }
>   scale 150*<4/3, 1, 4/3> }
> 
> This is a higher quality version. Set intervals to 1 and comment out samples
> for the version I posted above. If you want the whole scene let me know;
> you'll also need some suitable dxf files. They're about 9MB each, so I
> can't really post them!
> 
> Bill
> 
As the default method is method 3, or adaptive, it,s beter to only have 1 
interval, it's also faster.
The second value in samples is ignored as well.
Thus, if you want beter quality, you only need to increase samples.
In your case, it's beter and faster to use:
samples 300 without mentioning intervals whitch default to 1.
I've done some testings and samples 100 renders faster than interval 3 samples 
10 (only giving you 30 samples).

-- 
Alain
-------------------------------------------------
Bet my floppy's bigger than yours.

Post a reply to this message

Bruno Cabasson nous apporta ses lumieres en ce 25/08/2006 06:01:
> Hi there.
> 
> I recently worked on the question of clouds, and I found that one should
> separate cloud layers from individual clouds. I developed macros and
> template files in order to control atmposphere effects, cloud layers,
> individual clouds, redenning of light when the sun is low above the
> horizon.
> 
> One thing is sure: if you want nice clouds, it takes long to render because
> you need to increase the number of samples. You also must tune the ratio
> scatter/absorb, chose the scattering type (MIE_HAZY_SCATTERING seem OK),or
> maybe combine 2 scattering types. But I would not recomment adjusting the
> aspect via emission. First because it does not correspod to reality, and
> second because it will interact with radiosity in a wrong manner.
Only if you add "media on" in the radiosity block. Otherwise, radiosity will 
ignore your media, even if it's a very bright emissive one.
> 
> Here is my last result. I post right after some other samples of what I got
> so far. These are 800x600 renders with 2/3 of them is sky. They took 2-3
> hours ach with an athlon XP 3000+.
> 
> Btw: can I attach more than one image in a post?
Not when usint the web interface as in your current case.
Yes if using a mail/news client.
> 
> Now I am working on some other features: sun aspect (halo ...)
> 


-- 
Alain
-------------------------------------------------
Change is inevitable, except from vending machines.

Post a reply to this message

Alain <ele### [at] netscapenet> wrote:
> As the default method is method 3, or adaptive, it,s beter to only have 1
> interval, it's also faster.
> The second value in samples is ignored as well.
> Thus, if you want beter quality, you only need to increase samples.
> In your case, it's beter and faster to use:
> samples 300 without mentioning intervals whitch default to 1.
> I've done some testings and samples 100 renders faster than interval 3 samples
> 10 (only giving you 30 samples).

Nice one, thanks for the tip. I shall persevere. It's quite evident that
good clouds require a lot of work!

Bill

Post a reply to this message