Alain <aze### [at] qwertyorg> wrote:

> > nemesis<nam### [at] gmailcom>  wrote:
> >> how many months rendering? ;)
> >>
> >> Those clouds look way too good.  are they media-based?
> >
> > They use "df3" media in spheres.  Actually one "df3" file rotated randomly.
> > There are 90 clouds here, all copies of the same cloud.  Air is a Rayleigh
> > scattering media, with some emission to fake multiple scattering so it would not
> > be totally dark in the shaded areas.  Media used only 5 intervals, ( 1 to 4
> > samples (was it 2 to 8?)...more intervals would make a longer render (!!!!)
> >
> > Picture rendered at 1680 x 1050, antialiasing 0.1, radiosity on...only 5 or 10
> > samples...took about 4 to 6 hours to render on an eight core I7 with 64 bit
> > Windows 7 and 9 gigs of RAM.
> >
> >
> >
>
> What sampling method do you use? The default is method 3.
>
> When using method 3:
> intervals defaults to 1
> samples defaults to 10 (must be at least 3)
> Using intervals GREATLY increase the rendering time.
> The second samples value is ALWAYS ignored: samples 1,4 becomes samples 1
> intervals 10 samples 10 is MUCH slower than samples 100 (with default
> intervals 1).
>
> Expect you render tome to drop under 3 hours if you set only intervals 1
> and increase samples around 50. It's even possible that you could use
> something like samples 150 to 200 and still get faster rendering.
>
>
>
> Alain

Thank you for the tip -- will try that this weekend.  Will also refine some of
the inside plants, structures, etc, along with outside details (making the side
windows really transparent, which will slow render time by a factor of two or
so).

Ed

Post a reply to this message