Am 23.08.2010 00:09, schrieb Christian Froeschlin:
> clipka wrote:
>
>> The basic idea is pretty simple: Trace fewer radiosity sample rays if
>> they only hit non-important objects :-)
>
> Sounds great!
>
> How do you decide if an object is non-important?

I don't - the user does. At least that's how I've currently implemented 
it; e.g.:

   sphere {
     ...
     radiosity { importance 0.2 }
   }

> Would this also help for scenes where light falls
> through a narrow opening (door, blinds, ...)?

Should be possible, by placing a fully transparent object there with 
high importance.

Note however that this in turn means a glass windows will be sampled 
with the window pane's importance, unless the pane is marked as 
"no_radiosity" to fully hide it from sampling.

I might still come up with an even smarter idea. I guess the approach 
still leaves quite some room for further improvement.

Post a reply to this message

clipka <ano### [at] anonymousorg> wrote:
> The basic idea is pretty simple: Trace fewer radiosity sample rays if
> they only hit non-important objects :-)
>

Oh I don't know, I think the real test for radiosity is in how it handles "dark
corners".

Post a reply to this message