 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Just wow!
http://www6.incrysis.com/screenshots/27g4p.jpg
http://www6.incrysis.com/screenshots/67478845bf0.jpg
http://www6.incrysis.com/screenshots/palmspv2.jpg
http://www6.incrysis.com/screenshots/hrkduw.jpg
from: http://www6.incrysis.com/screenshots/27g4p.jpg
I wonder if there's still the need for slow academic CPU-bound rendering
methods, when the games industry has come up with such slightly less accurate,
but nonetheless amazing, approximations at a fraction of the cost.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> http://www6.incrysis.com/screenshots/27g4p.jpg
> http://www6.incrysis.com/screenshots/67478845bf0.jpg
> http://www6.incrysis.com/screenshots/palmspv2.jpg
> http://www6.incrysis.com/screenshots/hrkduw.jpg
Don't post photos! :-)
> I wonder if there's still the need for slow
Slow is an understatement, if doing a very accurate render of a similar
scene with path tracing or radiosity you are looking at 12 hours on a modern
CPU, that's 2 MILLION times slower than rendering in 1/60 of a second. The
quality/speed available from the GPU is awesome.
> academic CPU-bound rendering
> methods, when the games industry has come up with such slightly less
> accurate,
> but nonetheless amazing, approximations at a fraction of the cost.
Depends on the application, shots like you posted take *a lot* of work to
set up to be rendered on a GPU quickly, a lot of coding (shadows, shaders,
water effects, LOD, etc). This is of course ok for a game, but for
rendering a still image it will be quicker to just load the model into MCPov
and leave it for a day or two while you do something else :-)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
D***, this makes me want a high end SLI system :)
--
...Chambers
www.pacificwebguy.com
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> Don't post photos! :-)
Nah. If it was a photograph, it wouldn't have clearly visible polygon
edges in it. (E.g., the first image. The leaf in profile near the bottom
of the screen shows clear sharp corners rather than a smooth bend.)
Admittedly it's pretty damn impressive though.
> Depends on the application, shots like you posted take *a lot* of work
> to set up to be rendered on a GPU quickly, a lot of coding (shadows,
> shaders, water effects, LOD, etc). This is of course ok for a game, but
> for rendering a still image it will be quicker to just load the model
> into MCPov and leave it for a day or two while you do something else :-)
And also, I imagine, if you want to "accurately" simulate what, say, the
interior of a new building is going to look like.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> Depends on the application, shots like you posted take *a lot* of work to
>> set up to be rendered on a GPU quickly, a lot of coding (shadows,
>> shaders, water effects, LOD, etc). This is of course ok for a game, but
>> for rendering a still image it will be quicker to just load the model
>> into MCPov and leave it for a day or two while you do something else :-)
>
> And also, I imagine, if you want to "accurately" simulate what, say, the
> interior of a new building is going to look like.
Even for those animations of building fly throughs, you only run one
"lightmap" pass where all the lighting is accurately calculated and stored
in a file. The animation renderer then just references the pre-calculated
lighting while rendering each frame in a much faster way. Much like in a
game actually.
Unless you want an animation of how the building looks as the sun moves
across the sky, then you are in for a very long wait if you want perfectly
accurate results...
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
But isn't much of the illumination pre-generated on the texture bitmaps
through traditional slow methods? You can do that in POV-Ray too: render the
scene with a renderer that supports baking, put the resulting bitmap on the
illumination (see for instance
http://www.oyonale.com/modeles.php?lang=en&page=54).
Of course what they did in Crysis in much more complex, but I don't think
that they do GI in real time (though such stuff exists).
Also, CPU-bound rendering can be pretty nifty when it allows real time reuse
of illumination data, for instance
http://randomcontrol.com/index.php?option=com_content&view=article&id=61&Itemid=88
G.
web.49ab6b085273ea72e153145f0@news.povray.org...
> Just wow!
>
> http://www6.incrysis.com/screenshots/27g4p.jpg
> http://www6.incrysis.com/screenshots/67478845bf0.jpg
> http://www6.incrysis.com/screenshots/palmspv2.jpg
> http://www6.incrysis.com/screenshots/hrkduw.jpg
>
> from: http://www6.incrysis.com/screenshots/27g4p.jpg
>
> I wonder if there's still the need for slow academic CPU-bound rendering
> methods, when the games industry has come up with such slightly less
> accurate,
> but nonetheless amazing, approximations at a fraction of the cost.
>
>
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Gilles Tran wrote:
> Of course what they did in Crysis in much more complex, but I don't think
> that they do GI in real time (though such stuff exists).
According to Wikipedia, Crysis simulates GI in realtime using "ambient
occlusion", with no precomputed lightmaps.
(This is, however, not as accurate as a real GI pass.)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
49abd9f5$1@news.povray.org...
> Gilles Tran wrote:
>
>> Of course what they did in Crysis in much more complex, but I don't think
>> that they do GI in real time (though such stuff exists).
>
> According to Wikipedia, Crysis simulates GI in realtime using "ambient
> occlusion", with no precomputed lightmaps.
>
> (This is, however, not as accurate as a real GI pass.)
Thanks Andrew for the correction ("I don't think that.." = "I didn't bother
to do some minimal research but I'll write about it anyway...", grmphhh).
I see that there are some nice videos demonstrating the effect in real time.
http://video.google.com/videoplay?docid=-2592720445119800709&hl=en
It's true that an AO pass can be good enough in many situations, and games
are one of them. In product and architectural visualisation, AO is good for
the preliminary renderings but cannot replace GI for the final pics (though
a little AO can help with render times and overall realism). It's fast, so
people tend to overuse it and in the end it becomes annoying to see dirt in
every corner...
The real time subsurface scattering in Crysis is very impressive (from what
I see from stills and demos), I'm wondering whether it's only for flat
objects like leaves or if it works on actual volumes too.
I wish I had the time to play games, if only for the eye candy :(
G.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>>> Of course what they did in Crysis in much more complex, but I don't think
>>> that they do GI in real time (though such stuff exists).
>> According to Wikipedia, Crysis simulates GI in realtime using "ambient
>> occlusion", with no precomputed lightmaps.
>>
>> (This is, however, not as accurate as a real GI pass.)
>
> Thanks Andrew for the correction ("I don't think that.." = "I didn't bother
> to do some minimal research but I'll write about it anyway...", grmphhh).
Heh. It's only because I happened to read that article myself a few
weeks ago. ;-)
> It's true that an AO pass can be good enough in many situations, and games
> are one of them. In product and architectural visualisation, AO is good for
> the preliminary renderings but cannot replace GI for the final pics (though
> a little AO can help with render times and overall realism). It's fast, so
> people tend to overuse it and in the end it becomes annoying to see dirt in
> every corner...
Yes, well, what do you expect? ;-)
> The real time subsurface scattering in Crysis is very impressive (from what
> I see from stills and demos), I'm wondering whether it's only for flat
> objects like leaves or if it works on actual volumes too.
> I wish I had the time to play games, if only for the eye candy :(
It seems to work OK for the clouds at the start of the game - but then,
they could be faked some other way. It's part of a cutscene, so you
can't really check...
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Invisible" <voi### [at] dev null> wrote in message
news:49abf080@news.povray.org...
> It seems to work OK for the clouds at the start of the game - but then,
> they could be faked some other way. It's part of a cutscene, so you can't
> really check...
Have a play in the editor, open a Crysis level and look for the 'Time of
Day' settings wherever they are, and play with them, you'll be amazed at
what you can do. Don't save the level after changing the ToD though! ;)
~Steve~
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
