[MUD-Dev2] [DESIGN] Ray traced environments
John Buehler
johnbue at msn.com
Thu Mar 22 13:10:51 CET 2007
> John Buehler wrote:
> > Processors are fast. They're getting more numerous per box. We have
> > graphics coprocessors. Physics coprocessors are also out
> there. It may
> > be
> > time to start talking about raytracing in games.
> ... snip
>
>
> I have been saying this for awhile. Some random comments:
>
> - As the rendering technology gets more complex (such as ray tracing or
> stochastic ray tracing), GPUs lose their cost-efficient design
> compared to CPUs... which means that GPUs will eventually disappear, and
> you'll be using 12 cores from your 16 core processor to do rendering.
I assume the same things will happen.
> - Physics co-processors are still-born.
Agreed. Their existence suggests the desire for one box to do more. Too
bad for the PhysX guys that Cell and multi-core arrived at the same time.
> - Ray tracing doesn't give you that much over current scanline rendering,
> kind of. It provides slightly better shadows and curvy mirrors. It does,
> however, become very efficient when drawing LOTS of objects with LOTS of
> polygons, such as forests.
Yep. There's also the fact that a ray hits a "something", not just a
triangle. It increases the repertoire available to modellers. Purely
procedural objects spring to mind.
> The reason is that ray-tracing is O(pixels), while scan-line tends to be
> O(polygons) + a small contribution from O(pixels).
Spread the word :)
> - What does provide a huge improvement is stochastic ray tracing, which
> produces correct soft shadows, blurry reflections, diamonds, wine-glass
> effects, good motion blur, and secondary light bounces. It's 10x-1000x as
> slow as ray tracing though.
I'm not familiar with stochastic ray tracing, but given advances in
processing power, that should be on the table in another decade or two.
Desktop supercomputers should be able to handle that nicely.
JB
More information about the mud-dev2-archive
mailing list