|
|
Do you have "media on" in your radiosity block?
If so, it might be worthwhile tuning off radiosity->media interaction
and seeing if it still happens.
I've found that using scattering media with radiosity (and "media on")
to be prohibitively expensive and artifact prone, and have given up on
it. My time is better spent tweaking the scattering media to interact
properly with normal point light sources to get the desired effect.
Peter D.
Mike Thorn wrote:
> This is it: the horrendous image that I've been complaining about. I'm
> contemplating whether to change it and if so, exactly what to change
> (otherwise it gets junked directly). I've had a few ideas thrown at me
> from family members, so I might give it a shot.
>
> Anyway, I wanted to show this version to you all for opinions and
> comments. The WIP status indicates that it may undergo future drastic
> revamps, not its current development status.
>
> The thing in the middle is a big conical pool, which is what was sucking
> up so much render time. It rendered from the top of the image to the top
> edge of the pool in ten hours (including the pretrace), and has spent
> the last 122 hours 31 minutes and 8 seconds rendering the little bit
> from there on. Ugh.
>
> I'm particular interested in advice about the window spotlights. I've
> applied no density or turbulence to them and I'm getting a weird effect
> in the window media. It's like the windows don't have antialiasing on
> them. The lights are way out there (the room is 30 units wide and the
> lights are another 45 units out again) so there aren't any coincident
> surface problems, at least with the lights. Is there an obvious problem
> here or should I supply some code (not sure if that's legal under the
> rules).
>
> Thoughts and comments (or ideas) appreciated. I'm the worst critic this
> image has seen yet, so don't worry about hurting my feelings. :) I also
> apologize about the size - I felt it was necessary to be able to see the
> media detail.
>
> ~Mike
>
> ------------------------------------------------------------------------
>
|
|