> > let's hope that next generation of GPUs will prove even more versatile
> > allowing cleaner implementations (is random access to memory still so
> > far to come?).
>Yes definitly. Having normaly not random access is what makes gpu's fast.
>Look at data rates drop when using texture indirections. It seems things
>are not even going to improve with new memory technology, as penalties
>for band hopping are still tremendous.

Actually I had originally written "(is ra2m still a dream?)" :-)
But who knows what will those guys come up with? Its all evolving so
quickly... i.e. three years ago people laughed when one mentioned
raytracing on consumer hardware. (btw there are atleast 2 guys doing it
on PS2... thats what i call consumer! :-) if specs are those anticipated,
will we all use PS3s for GI? :)
Anyway already integer capabilities would be a nice thing, and that
shouldnt need any redesign...

>Btw if you mention Purcells Paper it is
>only fair to mention Carr et als, too they have a paper at the
>same conference with the same topic (GI strategies) on GPUs.

Uh thanks, had missed it, pretty nice!
Wanna see the thing live :-)

>Those times for radical different approaches are definitly over since
>years.

Since 1997? ;-)
We have seen that, but whats the reason? Dont say there's no place left to
explore anymore... it has been mostly scratching the surface so far.

>What I want to see more are spectral phenomens. (Yes they are
>HARD) How can someone call himself a Global Illumination solver without
>taking these effects into account.

Which ones do you consider hard?
I dont see much left to do on a theorical point of view...
...and most of such phenomena shouldnt bring substantial slowdown when
implemented carefully.
But as long as there are memory and bandwidth issues we'll all have to stick
with our silly color triplet. (btw i bet most commercial renderers still
implement
dispersion just by supersampling with rgbcolored rays/photons; anyone tested
Brazil and FRender to check it out?).
Instead I'd like to see more polarization around, but that's just an
implementation
issue, not that theres not enough material around...

\ Piero






>--
>Explore SRT with the help of Java3D
>(http://wwwvis.informatik.uni-stuttgart.de/relativity/minkowski)
>(http://www.antiflash.net/java3d/relativity (mirror)
>
>
>
>
>Community email addresses:
> Post message: realtime_raytracing@onelist.com
> Subscribe: realtime_raytracing-subscribe@onelist.com
> Unsubscribe: realtime_raytracing-unsubscribe@onelist.com
> List owner: realtime_raytracing-owner@onelist.com
>
>Shortcut URL to this page:
> http://www.onelist.com/community/realtime_raytracing
>
>Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/