 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
I've been improving the image the last month. This one is rendered in what
will correspond to 100dpi. If I can get posters printed at 200 or 300dpi
I'll render it at a higher resolution.
I changed the metal texture a lot and also its shape and welding. Does it
look right now?
I've worked on the stem too but I'm not satisfied with it. It looks like
it's made of metal, which is not the intention. It should look like a real
stem, only in black and white. It might help to just turn down the specular,
but there's also something odd about the small bumps that I can't figure
out. Maybe it needs a higher accuracy setting?
I tried to use subsurface scattering (sss) for the pistil and stamens but it
is behaving oddly. It creates bright pixels near the edge of the object, and
it also seem that blobs with sss look brigher than sphere_sweep with sss for
some reason. The bright end of the pistil is a blob while the rest is a
(straigt for now) sphere_sweep. I'm considering just dropping sss when it
creates so much trouble.
All comments, feedback and suggestions are welcome!
Rune
--
http://runevision.com
Post a reply to this message
Attachments:
Download 'compare_sss.jpg' (63 KB)
Download 'metal_a4_4_no_sss.jpg' (96 KB)
Preview of image 'compare_sss.jpg'
Preview of image 'metal_a4_4_no_sss.jpg'
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Rune" <new### [at] runevision com> wrote:
> I tried to use subsurface scattering (sss) for the pistil and stamens but it
> is behaving oddly. It creates bright pixels near the edge of the object, and
> it also seem that blobs with sss look brigher than sphere_sweep with sss for
> some reason. The bright end of the pistil is a blob while the rest is a
> (straigt for now) sphere_sweep. I'm considering just dropping sss when it
> creates so much trouble.
If you are using media for the SSS, then:
I don't know about sphere sweeps, but if I remember correctly the spurious
bright pixels around the edges of scattering media-filled blobs have been
described before somewhere in these NG. If I remember right, the problem
was intersection calculation inaccuracies in the blob object. A ray being
traced through the blob interior near its edge would fail to find the exit
intersection, and so the media code would decide it was facing a very long
ray and a correspondingly large amount of scattering medium -> brightness.
I *think* there was a solution; you specify the blob component sizes on a
larger scale, so instead of a spherical component with radius 0.1, you'd
use radius 10 and scale the blob down or the whole of the rest of the scene
up by a factor of 100 to compensate. I can't remember whether either option
(down-scaling or up-scaling) was preferred over the other. I also can't
remember whether a radius of 10 was sufficient to avoid the problems, but
I guess it would be easy to make a quick test scene and find out.
You might also fake SSS instead by using megapov's AOI pattern to add light
near the edges of the object, but I have never tried this.
Tom
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Rune nous apporta ses lumieres en ce 2007/07/27 15:05:
> I've been improving the image the last month. This one is rendered in what
> will correspond to 100dpi. If I can get posters printed at 200 or 300dpi
> I'll render it at a higher resolution.
>
> I changed the metal texture a lot and also its shape and welding. Does it
> look right now?
>
> I've worked on the stem too but I'm not satisfied with it. It looks like
> it's made of metal, which is not the intention. It should look like a real
> stem, only in black and white. It might help to just turn down the specular,
> but there's also something odd about the small bumps that I can't figure
> out. Maybe it needs a higher accuracy setting?
>
> I tried to use subsurface scattering (sss) for the pistil and stamens but it
> is behaving oddly. It creates bright pixels near the edge of the object, and
> it also seem that blobs with sss look brigher than sphere_sweep with sss for
> some reason. The bright end of the pistil is a blob while the rest is a
> (straigt for now) sphere_sweep. I'm considering just dropping sss when it
> creates so much trouble.
>
> All comments, feedback and suggestions are welcome!
>
> Rune
>
The weld look very good to me.
I loke the look of the reinforcing rings on the girder's holes. It's as if they
where stamped or riveted in place.
The stem looks somewhat metalic when in front of the white background, not when
there is something visible trough the transparence. Maybe adding some refraction
with an ior around that of water: ior 4/3
For the sss, maybe having a fade_color with a short fade_distance could help.
Does using strum help?
--
Alain
-------------------------------------------------
EVERYTHING HAS A GENDER
You may not know this but many nonliving things have a gender...
An Hourglass is Female, because over time, the weight shifts to the bottom.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Alain wrote:
> For the sss, maybe having a fade_color with a short fade_distance
> could help. Does using strum help?
Sturm in the blob makes no difference.
When I add fade_distance to the interior, things become even more odd: The
blob part seems not to become darker at all (and the artifacts are still
there). The sphere_sweep parts however become very dark (but the artifacts
are still here, too). It's the same interior applied to a union of blob and
sphere_sweep, yet they look completely different from each other! I don't
understand this...
Rune
--
http://runevision.com
Post a reply to this message
Attachments:
Download 'interior_oddness.jpg' (14 KB)
Preview of image 'interior_oddness.jpg'
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Tom York wrote:
> I *think* there was a solution; you specify the blob component sizes
> on a larger scale, so instead of a spherical component with radius
> 0.1, you'd use radius 10 and scale the blob down or the whole of the
> rest of the scene up by a factor of 100 to compensate.
Hmm. If I just make the blob components larger, the artifacts mostly
disappear. However, if I scale the whole blob down to compensate, they
reappear. I am reluctant to scaling my entire scene up at this point - it
would be an ugly hack. Thanks for the suggestion though.
> You might also fake SSS instead by using megapov's AOI pattern to add
> light near the edges of the object, but I have never tried this.
I don't think AOI can fake SSS except for very simple shapes like spheres,
cylinders and other completely convex shapes...? Even for those, it would
work only if the light comes almost from behind. It's the same reason you
can't use AOI to simulate shading of clouds. It's just not very related.
Rune
--
http://runevision.com
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Rune" <new### [at] runevisioncom> wrote:
> Alain wrote:
> > For the sss, maybe having a fade_color with a short fade_distance
> > could help. Does using strum help?
>
> Sturm in the blob makes no difference.
>
> When I add fade_distance to the interior, things become even more odd: The
> blob part seems not to become darker at all (and the artifacts are still
> there). The sphere_sweep parts however become very dark (but the artifacts
> are still here, too). It's the same interior applied to a union of blob and
> sphere_sweep, yet they look completely different from each other! I don't
> understand this...
>
> Rune
> --
> http://runevision.com
It could be about the angle of incidence of the light striking it. The
pistil has a lot of direct light, whereas the stamens don't (except the one
on the bottom where you can see a bright edge). I had some similar problems
with developing a wax material. I find it best not to mix interior fading
with media. Rather adjust the densities and colours of the scattering
media and use absorption media to attenuate the light instead. You may
also want to try other scattering media types than the one you have.
-tgq
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Maybe you should make the stem solid, I wouldn't expect
to see thru a stem of a plant, even if it's greyscale.
As for composition, have you considered making the
first part of the stem near the flower green-grey fading to grey?
That might give a visual clue that the stem is organic.
Or maybe blue-grey?
As it is, the stem reflects the flower, which makes that end
"hot", where you might expect a flower stem to be "cool".
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Tim Attwood wrote:
> Maybe you should make the stem solid, I wouldn't expect
> to see thru a stem of a plant, even if it's greyscale.
The transparency of the stem is not meant to be realistic. It's not shown in
this image, but the stem actually fades to completely transparent further
down, giving it an unreal feel. (The image shown here will be part of a
larger motive.)
> As for composition, have you considered making the
> first part of the stem near the flower green-grey fading to grey?
Yes, I did that initially. However, I found that the green color didn't fit
with the color scheme and ruined the style I was trying to create. So I have
decided to only have shades of gray and shades of red and orange in the
picture. Some real stems are red too anyway.
Rune
--
http://runevision.com
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Trevor G Quayle wrote:
> "Rune" wrote:
>> When I add fade_distance to the interior, things become even more
>> odd: The blob part seems not to become darker at all (and the
>> artifacts are still there). The sphere_sweep parts however become
>> very dark (but the artifacts are still here, too). It's the same
>> interior applied to a union of blob and sphere_sweep, yet they look
>> completely different from each other! I don't understand this...
>
> It could be about the angle of incidence of the light striking it.
> The pistil has a lot of direct light, whereas the stamens don't
> (except the one on the bottom where you can see a bright edge).
I found out it was just more precision problems. If both blob and
sphere_sweep are scaled up with a factor of ten, they are equally bright.
> I find it best not to mix interior fading with media.
Yes, I do use absorbing media. I only tested it with interior fading as per
Alain's suggestion.
I might play around a bit more and try to get the sss to work, but I found
some nice texture settings that look pretty nice too, so maybe sss won't be
necessary after all.
Rune
--
http://runevision.com
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Rune wrote:
> I might play around a bit more and try to get the sss to work, but I found
> some nice texture settings that look pretty nice too, so maybe sss won't be
> necessary after all.
>
> Rune
There's a certain sss technique I never posted here, but it may work
with this.
The idea is to scale a transparent copy of your target object slightly
larger than the visible object(with blobs, play with threshold). The
transparent object has a positive ior and refractive photons. The count
parameter in global_settings->photons adjusts how sharp the shadows
appear on your target object. Might be worth a try.
~Sam
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
