Stephen wrote:
> On Sun, 26 Oct 2008 00:18:31 +0300, Arttu Voutilainen
> <blizzara.REM0VE7H!S### [at] zbxtSP4MM3Rnet> wrote:
> 
>> But, before I render this in full size, I would like to hear any
>> comments and critics you guys have.
> 
> Your image is very attractive. The focal blur gives it a pointillist look. I
> look forward to seeing the final and fixed version.

Thanks!

-- Arttu Voutilainen

Post a reply to this message

Robert McGregor wrote:
>> I have no idea what causes the white lines on the sails. It is due to
>> fog-media, but I don't now why.
> 
> What are the sails made from? I've seen similar artifacts in my own work when
> using sphere_sweeps and other parametrics (which I've found is often due to
> rounding/precision errors). Scaling the entire scene several times larger often
> fixes that sort of thing for me. 

They are meshes (done in blender), but the problem was medias container
intersecting with the sails. It's fixed now.

> Okay, the critique: the near shoreline looks a bit too geometric (faceted) to me
> agaisnt the water, and I think the grass is too uniformly green.

Yep, I'll try to do something about those.

> Overall a nice scene though, very tranquil.


Thanks!

> -Rob

-- Arttu Voutilainen

Post a reply to this message

Kenneth wrote:
> Arttu Voutilainen <blizzara.REM0VE7H!S### [at] zbxtSP4MM3Rnet> wrote:
>> I have been working on this image since summer, and now, after 73 hours
>> of rendering, it's finally ready enough to be posted here.
> 
> The colors and lighting remind me of Maxfield Parrish; nicely done.

Thanks!

> I wish I could offer a suggestion about the 'lines,' but without seeing your
> scene code, it would be a shot in the dark.  Do you have any overlapping
> media-filled objects in the scene? Or is it just a single fog or atmospheric
> media?

There are actually three medias (one for atmosphere, one for clouds and
one for fog). It wasn't those though, just the fog-container
intersecting with the sails.

> Ken W.

-- Arttu Voutilainen

Post a reply to this message

Thomas de Groot wrote:
> Very nice composition indeed. In addition to some of the comments already 
> offered, I have doubts about the sun and the glare on the horizon. They seem 
> not to be in phase. I would suggest to take out the visible sun sphere.
> (I don't suppose it represents the Moon??  :-)  )
> 
Thanks! Yep, it's meant to be the sun, not moon. I guess I'll take it
off as now that you said it, I think it does look more like moon than sun :P

> Thomas 

-- Arttu Voutilainen

Post a reply to this message

Alain wrote:
> Arttu Voutilainen nous illumina en ce 2008-10-25 17:18 -->
>>
>> ------------------------------------------------------------------------
>>
> Try a test rander without focal blur nor radiosity.
> In that render, use a partly transparent texture for your media
> container. You may discover that those lines are where the media
> container intersect the sails.
> 
> If that's the case, try altering the container's shape or position so
> that it never intersect the sails. If you use a plane, just place it
> higher so that it's placed over the masts.
> 

That's it! Thanks!

> For your focal blur, try a smaller aperture.
> You may also do away with it and use a blured transparency just for the
> lower part. In that kind of scene, focal blur should be minimal, if
> present at all, and only visibly affect the extreme foreground.
> 

Yep, it actually turned out to be more blurred than I had expected. How
would I use blurred transparency?

-- Arttu Voutilainen

Post a reply to this message

William Tracy wrote:
> nemesis wrote:
>> Gee, man!  The render time was already 72 hours for quite low
>> resolution, let
>> alone with higher blur_samples.
> 
> Aren't I a bastard? ;-)
> 
> I look at the shoreline in the extreme foreground, and I see random blue
> and green pixels sprinkled around, which bugs me. With higher samples,
> you'll get more of a smooth gradient.
> 
> Another option would be to tweak the aperture, reducing the amount of
> blur, and reducing the number of samples needed to eliminate the grain.

Yep, I guess I'll do both: lower the aperture and increase samples. Or
if Alain can explain that blurred transparency to me well enough that I
can code it, I might try it first :P

> Then again, some people are fine with the grain. You decide whether the
> render time is worthwhile. :-)
> 
> BTW, now that you mention it, what hardware is this running on?
> 

The 72 hours was on my own computer, AMD 64 3500+. The bigger version
(now 140:57:56 Rendering line 130 of 720, 89598 rad. samples) is running
on Eero Ahonen's bit more faster AMD 4450e.

-- Arttu Voutilainen

Post a reply to this message

Kirk Andrews wrote:
> Hi Arttu,
> 
> Very nice image.  I love sunset scenes, and I like the mood and subject of your
> image.  You've done really well with media.

Thanks!

> That said, I offer a few things that I would do:
> 
> With such atrocious render times, I would actually drop the media--there's
> nothing in this image that could not be achieved with a good skysphere and fog.
>  In fact, a good skysphere could probably even offer a more realistic and
> dramatic result.

I guess it would indeed work. It's just that this whole scene evolved
from my atmosphere-experiments (which were mostly inspired by your
atmospheres and clouds, actually ;)) and it's nice to know that I can
chance the whole lightning by just moving my sun around. It's another
question if that's worth the rendering times, but :P


> I would also drop the focal blur.  I agree with the others--focal blur in this
> image would be minimal, and it would be easier to simply add a little blur to
> the foreground later in GIMP or Photoshop.  I think you'll find that focal blur
> is the cause of the majority of your render time.

Yep, at least I have rendered this in full size once before w/o blur,
and it took only 'bout 14 days ;P But yeah, I'll see what I'll do about
the blur.

> I think your greens are too green and too homogeneous.  My recommendation would
> be to drop it down to something like rgb <.1,.2,.0>, but use a cells pigment
> with turbulence 1, lambda 10 to get some nice, realistic variation.  I would
> apply this to both the tree leaves and the grass.  I know you mentioned grass
> being in the render, but I can's see it (perhaps because the focal blur is
> obscuring it).  If it is there, I would make it much taller and more irregular.
>  For the foreground, I think you need to add some quite tall grass and
> vegetation.  Gilles Tran has several varieties of grass-like plants that you
> could add.

There really is real grass (Gilles' MakeGrass) in foreground, but the
focal blur ate it away :(. I'm experimenting with that cells pattern
right now, and it looks promising. Plants sounds like a good idea; I'll
try those too.


> I don't know what method you used to create the terrain, but I think adding some
> more irregularity and character would add a lot.  When I create terrain, I
> usually use an isosurface or a pattern-generated heightfield using a function
> based on something like this:
> 
> function {
>   pattern {
>     wrinkles
>     poly_wave 3
>   }
> }
> 
> Or, you could start with the terrain you have, but use trace to drop in
> isosurface rocks and boulders along the banks of the river to get a bank that
> doesn't simply slope straight into the water.

It's a heighfield which is generated from an image that is made with The
Gimp. I'll try to do something with it. Also the idea about rocks sounds
good.

And thanks for the function, I have always wondered how you make your
landscapes. I don't think I'll use it in this image though, but I'll
remember it next time when I'm doing landscapes :)

> Those are my suggestions; I hope you find them helpful--I look forward to seeing
> what you come up with.

Thanks a lot, I really appreciate them. I'll see what I can do.

> Kirk

-- Arttu Voutilainen

Post a reply to this message

Cousin Ricky wrote:
> Arttu Voutilainen <blizzara.REM0VE7H!S### [at] zbxtSP4MM3Rnet> wrote:
>> I have been working on this image since summer, and now, after 73 hours
>> of rendering, it's finally ready enough to be posted here.
> 
> It looks like a painting combining the styles of Georges Seurat and the old
> masters.

Thanks.

> A couple of criticisms: the Sun should be the brightest object in the scene, and
> the clouds near the Sun's line of sight should have brighter fringes.

Yep, you are right. That will be fixed.

-- Arttu Voutilainen

Post a reply to this message

Arttu Voutilainen nous illumina en ce 2008-10-27 13:58 -->
> William Tracy wrote:
>> nemesis wrote:
>>> Gee, man!  The render time was already 72 hours for quite low
>>> resolution, let
>>> alone with higher blur_samples.
>> Aren't I a bastard? ;-)
>>
>> I look at the shoreline in the extreme foreground, and I see random blue
>> and green pixels sprinkled around, which bugs me. With higher samples,
>> you'll get more of a smooth gradient.
>>
>> Another option would be to tweak the aperture, reducing the amount of
>> blur, and reducing the number of samples needed to eliminate the grain.
> 
> Yep, I guess I'll do both: lower the aperture and increase samples. Or
> if Alain can explain that blurred transparency to me well enough that I
> can code it, I might try it first :P
> 
>> Then again, some people are fine with the grain. You decide whether the
>> render time is worthwhile. :-)
>>
>> BTW, now that you mention it, what hardware is this running on?
>>
> 
> The 72 hours was on my own computer, AMD 64 3500+. The bigger version
> (now 140:57:56 Rendering line 130 of 720, 89598 rad. samples) is running
> on Eero Ahonen's bit more faster AMD 4450e.
> 
> -- Arttu Voutilainen

Have you ever done any blured reflection?
Now, do the same, but with a transparent object placed close to the camera.
The thing is, you need to keep the number of averaged normals low, as each ray 
will hit the surfaces twice. Averaging 16 normale résult in each ray been 
multiplied by 16x16, or 256.


-- 
Alain
-------------------------------------------------
I bet exercise equipment would be a lot more expensive if we had evolved from 
starfish.

Post a reply to this message

Alain wrote:
> Arttu Voutilainen nous illumina en ce 2008-10-27 13:58 -->
>> William Tracy wrote:
>>> nemesis wrote:
>>>> Gee, man!  The render time was already 72 hours for quite low
>>>> resolution, let
>>>> alone with higher blur_samples.
>>> Aren't I a bastard? ;-)
>>>
>>> I look at the shoreline in the extreme foreground, and I see random blue
>>> and green pixels sprinkled around, which bugs me. With higher samples,
>>> you'll get more of a smooth gradient.
>>>
>>> Another option would be to tweak the aperture, reducing the amount of
>>> blur, and reducing the number of samples needed to eliminate the grain.
>>
>> Yep, I guess I'll do both: lower the aperture and increase samples. Or
>> if Alain can explain that blurred transparency to me well enough that I
>> can code it, I might try it first :P
>>
>>> Then again, some people are fine with the grain. You decide whether the
>>> render time is worthwhile. :-)
>>>
>>> BTW, now that you mention it, what hardware is this running on?
>>>
>>
>> The 72 hours was on my own computer, AMD 64 3500+. The bigger version
>> (now 140:57:56 Rendering line 130 of 720, 89598 rad. samples) is running
>> on Eero Ahonen's bit more faster AMD 4450e.
>>
>> -- Arttu Voutilainen
> 
> Have you ever done any blured reflection?
> Now, do the same, but with a transparent object placed close to the camera.
> The thing is, you need to keep the number of averaged normals low, as
> each ray will hit the surfaces twice. Averaging 16 normale résult in
> each ray been multiplied by 16x16, or 256.
> 
> 

Nope, I have never really done it, but I think I tried once. I do know
how it's supposed to be done, and I'll try it when I have time to get
back to this image. School has kept me a bit busy lately..

Thanks for the tip.

-- Arttu Voutilainen

Post a reply to this message