 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
I've made a smooth landscape using an isosurface containing noise3d + y
. Note in the image below how some of the features are truncated: flat
tops. I want it to be completely smooth, but have this general feel. I
was trying to figure out what might have been wrong with my code. But
then I remembered a similar thread about some of the noise-based
pigments in povray. Did I hear something to the effect that the pov
designers used an unorthodox definition of the noise (something about
the interval?) ?
The question is: which of the scenarios below are most probable:
1. I may acheive smoothness by fixing my code;
2. I will fail because these flat tops are inherent in any (correct)
application of noise;
3. I will fail until a pov version comes out with a "correct"
application of noise.
Post a reply to this message
Attachments:
Download 'isosu368.jpg' (5 KB)
Preview of image 'isosu368.jpg'
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Fri, 07 Jan 2000 08:31:44 -0500, Greg M. Johnson wrote:
>The question is: which of the scenarios below are most probable:
>1. I may acheive smoothness by fixing my code;
>2. I will fail because these flat tops are inherent in any (correct)
>application of noise;
>3. I will fail until a pov version comes out with a "correct"
>application of noise.
My guess is 1. I don't think noise3d suffers from the same problems
as bozo and other patterns. Try making your bounding box larger.
--
These are my opinions. I do NOT speak for the POV-Team.
The superpatch: http://www2.fwi.com/~parkerr/superpatch/
My other stuff: http://www2.fwi.com/~parkerr/traces.html
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Thanks for the speedy reply.
No, I believe I have a bounding SPHERE, but I'll have to check the code
at home this evening.
I acheived this surface by something on the order of:
isosurface{noise3d(x,y,z) + y }.
___________________________
Greg M. Johnson http://members.xoom.com/gregjohn/animation.html
"Doesn't raytrace at work!"
Ron Parker wrote:
> On Fri, 07 Jan 2000 08:31:44 -0500, Greg M. Johnson wrote:
>
> >The question is: which of the scenarios below are most probable:
> >1. I may acheive smoothness by fixing my code;
> >2. I will fail because these flat tops are inherent in any (correct)
> >application of noise;
> >3. I will fail until a pov version comes out with a "correct"
> >application of noise.
>
> My guess is 1. I don't think noise3d suffers from the same problems
> as bozo and other patterns. Try making your bounding box larger.
>
> --
> These are my opinions. I do NOT speak for the POV-Team.
> The superpatch: http://www2.fwi.com/~parkerr/superpatch/
> My other stuff: http://www2.fwi.com/~parkerr/traces.html
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
In article <3875ec85@news.povray.org>, ron### [at] povray org wrote:
> My guess is 1. I don't think noise3d suffers from the same problems
> as bozo and other patterns. Try making your bounding box larger.
Actually, it does, but it doesn't show up in most images because the
flat area is in the center of the blobs formed. This simple isosurface
demonstrates the problem:
isosurface {
function {y - noise3d(x*2, 0, z*2)*0.33}
eval
max_gradient 1.1
threshold 0
bounded_by {box {<-2,-2,-2>, < 2, 2, 2>}}
accuracy 0.001
texture {
pigment {color rgb < 0.095, 0.45, 0.2>}
}
}
--
Chris Huff
e-mail: chr### [at] yahoocom
Web page: http://chrishuff.dhs.org/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
I may be compeletely wrong, because I don't have a lot of experience
with isosurfaces, but I did run into something similar. It seems to me
that the isosurface code works "backwards" from normal POV-Ray
nomenclature. i.e. something like isosurface{y-noise3d(x,y,z)} actually
adds the height of the noise to the plane whereas
isosurface{y+noise3d(x,y,z)} subtracts the noise from the surface.
Likewise it seems that increasing the x,y,and z values makes for smaller
noise features. e.g. noise3d(x*20,y*20,z*20) makes finer detail than
noise3d(x*3,y*3,z*3).
You may simply be running into your bounding box. Try translating the y
component of your isosurface down a bit at see it the tops of your noise
"finish out".
Mike Wilson
"Greg M. Johnson" wrote:
>
> I've made a smooth landscape using an isosurface containing noise3d + y
> . Note in the image below how some of the features are truncated: flat
> tops. I want it to be completely smooth, but have this general feel. I
> was trying to figure out what might have been wrong with my code. But
> then I remembered a similar thread about some of the noise-based
> pigments in povray. Did I hear something to the effect that the pov
> designers used an unorthodox definition of the noise (something about
> the interval?) ?
>
> The question is: which of the scenarios below are most probable:
> 1. I may acheive smoothness by fixing my code;
> 2. I will fail because these flat tops are inherent in any (correct)
> application of noise;
> 3. I will fail until a pov version comes out with a "correct"
> application of noise.
>
> ------------------------------------------------------------------------
> [Image]
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
In article <3875FB3E.46031B75@iastate.edu>, maw### [at] isunetnet wrote:
> I may be compeletely wrong, because I don't have a lot of experience
> with isosurfaces, but I did run into something similar. It seems to me
> that the isosurface code works "backwards" from normal POV-Ray
> nomenclature. i.e. something like isosurface{y-noise3d(x,y,z)} actually
> adds the height of the noise to the plane whereas
> isosurface{y+noise3d(x,y,z)} subtracts the noise from the surface.
That isn't backwards at all. The isosurface of the function y-1 will
consist of all points where y-1 is the threshold value. Increasing the
amount subtracted from the y coordinate means y has to be higher to
result in the same value. When noise3d equals 1, y has to equal 1. When
it is added instead, y has to equal -1 to get the same result.
> Likewise it seems that increasing the x,y,and z values makes for smaller
> noise features. e.g. noise3d(x*20,y*20,z*20) makes finer detail than
> noise3d(x*3,y*3,z*3).
Again, this is not backwards. If you multiply x by 2, than the result of
that expression(2*x) will be a certain value at half the distance it
would be without the multiplication. The result is compressing the
coordinates into a smaller space, scaling down.
If x is 3, 2*x will be 6, and the features that would have been at x=6
will be at x=3 when 2*x is used.
--
Chris Huff
e-mail: chr### [at] yahoocom
Web page: http://chrishuff.dhs.org/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
You're right. of course. I wasn't implying that it was wrong, just that
it "appears" backwards compared to other types of scaling. For example,
if I wanted a large bump scale on a normal, I would increase it, but
with noise3d, I would use a smaller number.
As, I mentioned, I don't have a lot of experience with isosurfaces, so I
don't think in terms of how the function will be satisfied; I'm relating
it to similar POV-Ray statements.
Mike Wilson
Chris Huff wrote:
>
> In article <3875FB3E.46031B75@iastate.edu>, maw### [at] isunetnet wrote:
>
> > I may be compeletely wrong, because I don't have a lot of experience
> > with isosurfaces, but I did run into something similar. It seems to me
> > that the isosurface code works "backwards" from normal POV-Ray
> > nomenclature. i.e. something like isosurface{y-noise3d(x,y,z)} actually
> > adds the height of the noise to the plane whereas
> > isosurface{y+noise3d(x,y,z)} subtracts the noise from the surface.
>
> That isn't backwards at all. The isosurface of the function y-1 will
> consist of all points where y-1 is the threshold value. Increasing the
> amount subtracted from the y coordinate means y has to be higher to
> result in the same value. When noise3d equals 1, y has to equal 1. When
> it is added instead, y has to equal -1 to get the same result.
>
> > Likewise it seems that increasing the x,y,and z values makes for smaller
> > noise features. e.g. noise3d(x*20,y*20,z*20) makes finer detail than
> > noise3d(x*3,y*3,z*3).
>
> Again, this is not backwards. If you multiply x by 2, than the result of
> that expression(2*x) will be a certain value at half the distance it
> would be without the multiplication. The result is compressing the
> coordinates into a smaller space, scaling down.
> If x is 3, 2*x will be 6, and the features that would have been at x=6
> will be at x=3 when 2*x is used.
>
> --
> Chris Huff
> e-mail: chr### [at] yahoocom
> Web page: http://chrishuff.dhs.org/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> If x is 3, 2*x will be 6, and the features that would have been at x=6
> will be at x=3 when 2*x is used.
>
Whoa, did I not catch on to this reasoning at all!
If x=3 it will be x=3 when 2*x??
Okay, I'm lost. Unless it's a average of the noise function....
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
In article <38764dcf@news.povray.org>, "omniVERSE" <inv### [at] aolcom>
wrote:
> Whoa, did I not catch on to this reasoning at all!
> If x=3 it will be x=3 when 2*x??
> Okay, I'm lost. Unless it's a average of the noise function....
No, if the function returns a certain value at 6 units along the x
axis(x=6), than if it is given 2*x instead of x, it will think x=6 when
it is really x=3. So the value that showed up at 6 units now shows up at
3 units.
N=What the function "sees"
V=The real value
With N=V:
0 1 2 3 4 5 6
N | | | | | | |
0 1 2 3 4 5 6
V | | | | | | |
With N=2*V
0 1 2 3 4 5 6
N | | | | | | |
0 1 2 3 4 5 6
V | | | | | | |
See what I mean now? If the function returned A at x=6 when given x, it
will give A at x=3 when given 2*x.
Sorry, I am not good at explaining this kind of thing.
--
Chris Huff
e-mail: chr### [at] yahoocom
Web page: http://chrishuff.dhs.org/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Well, you're talking to an ignoramus here anyway, so that doesn't help at
all.
The text spacing suffered in the translation but I guess I get you're point
anyhow. What's perplexing is why the "x" ever changes if I'm understanding
this whatsoever. Idea is that it gets seen together with whatever equation
then? x=6 is also (2*x)=3 just simply sounds like division, (x/2)=3, in the
usual sense. Maybe that's where other people are going wrong in the
thinking.
Oh! Do you mean that "x" is the only thing checked in some way? So despite
the resulting math it looks back at "x" alone??
Maybe I'm catching on, but you need not try and explain further unless
someone else is trying figuring this stuff out.
Bob
"Chris Huff" <chr### [at] yahoocom> wrote in message
news:chrishuff_99-D08175.16055207012000@news.povray.org...
> In article <38764dcf@news.povray.org>, "omniVERSE" <inv### [at] aolcom>
> wrote:
>
> > Whoa, did I not catch on to this reasoning at all!
> > If x=3 it will be x=3 when 2*x??
> > Okay, I'm lost. Unless it's a average of the noise function....
>
> No, if the function returns a certain value at 6 units along the x
> axis(x=6), than if it is given 2*x instead of x, it will think x=6 when
> it is really x=3. So the value that showed up at 6 units now shows up at
> 3 units.
>
> N=What the function "sees"
> V=The real value
>
> With N=V:
>
> 0 1 2 3 4 5 6
> N | | | | | | |
>
> 0 1 2 3 4 5 6
> V | | | | | | |
>
>
> With N=2*V
>
> 0 1 2 3 4 5 6
> N | | | | | | |
>
> 0 1 2 3 4 5 6
> V | | | | | | |
>
> See what I mean now? If the function returned A at x=6 when given x, it
> will give A at x=3 when given 2*x.
>
> Sorry, I am not good at explaining this kind of thing.
>
> --
> Chris Huff
> e-mail: chr### [at] yahoocom
> Web page: http://chrishuff.dhs.org/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
