 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Let's use the GPU to speed up POV-Ray!
http://www.cgarchitect.com/news/SIGGRAPH-2009-CHAOS-GROUP-GPU.shtml
</Saul>
--
Darren New, San Diego CA, USA (PST)
Understanding the structure of the universe
via religion is like understanding the
structure of computers via Tron.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Darren New schrieb:
> Let's use the GPU to speed up POV-Ray!
>
> http://www.cgarchitect.com/news/SIGGRAPH-2009-CHAOS-GROUP-GPU.shtml
Quite impressive. Seems better suited to speed up MCPov than POV-Ray
though :-P
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Actually not bad, for having a single primitive type and an extremely
simplified test scene :)
BTW, one thing I've always disliked about many GI renderers is the need
to perform multiple passes to get rid of the speckles. POV's method of
a single, high quality pass is, I believe, much to be preferred.
...Chambers
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Chambers wrote:
> Actually not bad, for having a single primitive type and an extremely
> simplified test scene :)
The test scenes later in the video (I admit it's tedious to watch it
straight thru) are quite complex.
--
Darren New, San Diego CA, USA (PST)
Understanding the structure of the universe
via religion is like understanding the
structure of computers via Tron.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Chambers schrieb:
> Actually not bad, for having a single primitive type and an extremely
> simplified test scene :)
>
> BTW, one thing I've always disliked about many GI renderers is the need
> to perform multiple passes to get rid of the speckles. POV's method of
> a single, high quality pass is, I believe, much to be preferred.
For real-time rendering the multi-pass approach is much better suited
though: Just re-starting the render whenever the user pans or dollys
around gives you a simple and low-latency way of trading speed vs. quality.
I don't think it matters much /how/ a good-quality render is achieved,
provided that it is achieved in due time. I'd actually prefer a coarse
speckled preview over being able to quickly see the top-left corner in
full quality :-), and the only reason why I prefer POV-Ray over MCPov is
that I think POV-Ray in most cases gives almost the same quality in much
shorter time (due to the caching of data), and that I find MCPov hard to
set up (due to the inevitable hassle with portals because it doesn't
support POV-Ray's classic lighting model).
And strictly speaking, POV-Ray can't do GI in a single pass either: It
needs radiosity pretrace to achieve any reasonable quality.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Darren New <dne### [at] san rrcom> wrote:
> Let's use the GPU to speed up POV-Ray!
>
> http://www.cgarchitect.com/news/SIGGRAPH-2009-CHAOS-GROUP-GPU.shtml
Completely mindblowing, specially the later scenes! o_O That's why V-Ray is top
today, without even resorting to unbiased algorithms.
I can also see real-time raytracing in nextgen game engines sooner than
expected.
> </Saul>
huh?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
nemesis wrote:
>> </Saul>
> huh?
An indication that I wasn't being serious about POV-Ray implementing it.
--
Darren New, San Diego CA, USA (PST)
Understanding the structure of the universe
via religion is like understanding the
structure of computers via Tron.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
nemesis schrieb:
> Completely mindblowing, specially the later scenes! o_O That's why V-Ray is top
> today, without even resorting to unbiased algorithms.
Sure it's unbiased? Looks like a monte-carlo approach to me.
>
> I can also see real-time raytracing in nextgen game engines sooner than
> expected.
Yup. Though gamers will need to get used to the noisy images when they
move... which is virtually constantly :-P.
The game engine designers will also need to find a way to not fully
re-render a scene just because /something/ in it moved.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
clipka wrote:
> Yup. Though gamers will need to get used to the noisy images when they
> move... which is virtually constantly :-P.
There's already noise in images. It just shows up as a lack of detail,
instead. (Of course, you'd probably have both, since they're coming from
different causes.)
> The game engine designers will also need to find a way to not fully
> re-render a scene just because /something/ in it moved.
I wouldn't think that's difficult, if you're doing GI to start with.
Certainly only the stuff you didn't see before would need to be re-rendered.
--
Darren New, San Diego CA, USA (PST)
Understanding the structure of the universe
via religion is like understanding the
structure of computers via Tron.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Darren New schrieb:
>> The game engine designers will also need to find a way to not fully
>> re-render a scene just because /something/ in it moved.
>
> I wouldn't think that's difficult, if you're doing GI to start with.
> Certainly only the stuff you didn't see before would need to be
> re-rendered.
... and that space where /something/ was before.
And anything that could possibly be affected by the resulting change in GI.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
