Before I post a bug report, I'd like some confirmation of what I've found. There
seems to be a (very conspicuous) bug in the various noise functions. It exhibits
itself as banding after a certain _negative_ distance, along any axis, from the
origin. This distance varies depending on the noise function: it is around
78.125 for 1/f noise (granite), 312.5 (78.125*4) for DNoise and Agate noise, and
around 10000 for Perlin noise (bozo, bumps, spotted).
Furthermore, in 1/f and DNoise the bug seems to become more pronounced at
78.125*2 and 78.125*4
I've tested this on POVWin 3.1g (Watcom and VC compile) and the Superpatch (3.1e
codebase)

//BEGIN EXAMPLE SCENE

camera{
    orthographic
    location y*5
    up y*5*.75 right x*5
    look_at 0
}

//1/f noise (granite)
#declare Tex1=
pigment{
    granite
    color_map{[0 rgb 0][1 rgb 1]}
    translate 78.13*x //First signs of banding
    //translate 156.25*x //More pronounced
    //translate 312.5*x //Even more pronounced
    scale 5
}

//Perlin noise (Bozo & Bumps & Spotted)
#declare Tex2=
pigment{
    bozo
    color_map{[0 rgb 0][1 rgb 1]}
    translate 10000*x
    scale 1
}

//DNoise (turbulence)
#declare Tex3=
pigment{
    gradient x
    color_map{[0 rgb 0][1 rgb 1]}
    translate 312.5*x //Banding starts
    //translate 625*x //Becomes more pronounced
    scale 1
    turbulence 1
}

//Agate turbulence
#declare Tex4=
pigment{
    agate
    color_map{[0 rgb 0][1 rgb 1]}
    translate 312.5*x //Banding starts
    scale 5
}


plane{
    y,0
    pigment{Tex1}
    finish{ambient 1}
}

Post a reply to this message

Margus Ramst wrote:
> 
> Before I post a bug report, I'd like some confirmation of what I've found. There
> seems to be a (very conspicuous) bug in the various noise functions. It exhibits
> itself as banding after a certain _negative_ distance, along any axis, from the
> origin. This distance varies depending on the noise function: it is around
> 78.125 for 1/f noise (granite), 312.5 (78.125*4) for DNoise and Agate noise, and
> around 10000 for Perlin noise (bozo, bumps, spotted).
> Furthermore, in 1/f and DNoise the bug seems to become more pronounced at
> 78.125*2 and 78.125*4
> I've tested this on POVWin 3.1g (Watcom and VC compile) and the Superpatch (3.1e
> codebase)

I see what you see I think:

1- tex1 hardly noticable but I see two horizontal bands at 1/3 and 2/3
2- tex2 Very pronounced on left side of screen
3- tex3 Unsure but possibly similar to #1
4- tex4 Down right freaky - prevelant in upper left of image

 For what it's worth they render with the same results in Pov v3.01r1.watcom
so if it is a bug or programming error it has been around for quite a while.
When I get time I will reformat your code and check it with my install of
Pov v2.2 and see if it goes back even further.

-- 
Ken Tyler

See my 850+ Povray and 3D Rendering and Raytracing Links at:
http://home.pacbell.net/tylereng/index.html

Post a reply to this message

Ken wrote:

> When I get time I will reformat your code and check it with my install of
> Pov v2.2 and see if it goes back even further.

Forget that. There is no orthographic camera type in v2.2 so the results will
not be similar regardless.

-- 
Ken Tyler

See my 850+ Povray and 3D Rendering and Raytracing Links at:
http://home.pacbell.net/tylereng/index.html

Post a reply to this message

Try the standard camera, results are the same.

Margus

Ken wrote:
> 
> Ken wrote:
> 
> > When I get time I will reformat your code and check it with my install of
> > Pov v2.2 and see if it goes back even further.
> 
> Forget that. There is no orthographic camera type in v2.2 so the results will
> not be similar regardless.
> 
> --
> Ken Tyler
> 
> See my 850+ Povray and 3D Rendering and Raytracing Links at:
> http://home.pacbell.net/tylereng/index.html

Post a reply to this message

Margus Ramst wrote:
> 
> Before I post a bug report, I'd like some confirmation of what I've found. There
> seems to be a (very conspicuous) bug in the various noise functions. It exhibits
> itself as banding after a certain _negative_ distance

You are likely seeing roundoff errors.  The basic solid noise function
is only good down to around -10000.  After an octave or two (chop in
half each time), this number gets pretty small.  Could be the reason you
are seeing it at -80 or so.

To test this theory, you can change the definition of MINX (in
texture.c) to something along the lines of -10000000 (plenty of digits
in a double to handle six or seven digits in the integer part and a
whole bunch below).

I've seen this sort of thing before.  In that particular instance, the
coordinates of the thing I was rendering were very small (like -100000),
with the camera down there also.  Turned bumps into a really odd checker
sort of thing.  Changing coordinates did the trick.

Xander

Post a reply to this message