"Dre" <and### [at] gmailcom> wrote:
> "fidos" <fid### [at] wanadoofr> wrote in message
> news:web.498609b1334f43c14f44ad660@news.povray.org...
> > "triple_r" <nomail@nomail> wrote:
> >> "fidos" <fid### [at] wanadoofr> wrote:
> >> > 3 hours with mcpov for a RSOCP scene.
> >>
> >> Neat.  Three hours, eh?  I'll assume you need portals to get it to render
> >> that
> >> quickly, even on a quad-core.  How do you set that up for toroidal lights
> >> though?  Do you bound it by a sphere?  Do you approximate the tori with a
> >> mesh?
> >>  Or maybe you have other tricks up your sleeve...
> >>
> >>  - Ricky
> >
> > No portal. There is too much light sources and portals would be too slow.
> > I use
> > high number of primary diffuse rays (12).
> >
> > Attached the source of the scene.
> >
> > Regards,
> > Fidos.
> >
> Can I ask how many passes this scenes took?
>
> I'm just curious as I have it running atm and its on pass 14 and still very
> very blurry.  I dont have the CPU power you do but its been going to for
> twice as long allready.
>
> Cheers Dre

Around 28 passes on 4 cores if I remember well. But, 28x8 passes is not
equivalent to 56x2 or 112x1 passes. Better is 28x8 passes.
H. Karsten point out that in its post :
http://news.povray.org/web.49563c4fbe8b348c20776f0%40news.povray.org
The problem is in the way mcpov choise the pixels to trace in order to reduce
noise.

Regards,
Fidos

Post a reply to this message