"Slime" <fak### [at] emailaddress> wrote:
> Looks pretty good.
>
> > For some reason this took eons
> > to render, something in the order of 5 days (partially because of 256
> light
> > sources and AA 0.0).
>
> Why did you need that many light sources? It looks like a few area lights
> and *maybe* some radiosity would have done fine.
>
>  - Slime
>  [ http://www.slimeland.com/ ]

Yes, it needs that may light sources.  The lighting is the HDRI-based light
system I've been perpetually working on.  It uses median-cut method (plus
some other stuff) so it is a n^2 problem.  I usually use a level of 8 to
get the smoothest minimum results (i.e. 2^8=256 sources).  Just simply
placing area lights wouldn't give a 'proper' light distribution for the
scene, as it calculates light from the entire HDRI environment map.  Unlike
radiosity, it uses 'active' light sources.  I can't get as good results with
radiosity without extremely high settings.
Usually the scenes don't run as long as this one, I can't figure out what
slows it down so much.  It appears to come from the csg gears mostly.  I
did get it to run somewhat better using manual bounding.
As, I've stated previously, I hope to post my lighting system when I'm fully
satisfied with it's performance.

-tgq

Post a reply to this message