"Thomas de Groot" <t.d### [at] internlDOTnet> wrote:
> "Bruno Cabasson" <bru### [at] alcatelaleniaspacefr> schreef in bericht
> news:web.46496bd58b4e344ab227a3bb0@news.povray.org...
> > Hello folks!
> >
> > Experimenting lately with atmospheres and clouscapes, I got this render
> > quite satisfactory, though you can notice some little under-sampling
> > artifacts. Instead of faking the Nature, I tried to follow the laws of
> > physics and let them do the job.
> >
> > The ground is basic and has a white pigment. The light is white. This
> > render
> > does not use radiosity, thus all turns very red, but it is a normal
> > behaviour. With radiosity, I guess the red color coming from the white sun
> > (after Rayleigh scattering of the atmosphre) would participate less to
> > illumination, and the lighting of objects would be more realistic. I'll
> > experiment soon with radiosity on those scenes.
> >
> > Comments welcome.
> >
>
> I am really impressed by this. Very natural looking. In the line of Abe
> Madey's work. How long did the render take?
>
> Thomas

For this 1024*768 render, it took a bit more than one day on my Athlon
XP3000+. But most of the image is occupied by the clouds. I noticed that
with clouds, the  finest the resolution, the more samples you need. Here I
used 60 samples for 1 interval and method 3 for Mie scattering. I think I
should have set it up to 100 to avoid some artifacts when the media is thin
(sampling must 'hit' the media) or near the horizon with low sun (the light
makes a long path through the container - differenced spheres).

I am currently rendering a smaller 512*384 with 2 layers of clouds and a
more red light. I'll post the result tomorrow if you are interrested. I am
focusing on sunsets because they are much more difficult than daylight
clouds: needs more samples due to the longer distance the light traverses.

    Bruno

Post a reply to this message