// Persistence of Vision Ray Tracer Scene Description File
// File: IsoDessert.pov
// Vers: 3.5
// Desc: Tests using isosurfaces for landscapes
// Date: 2/15/2005
// Auth: Bryan Valencia

// ==== Standard POV-Ray Includes ====
#include "colors.inc"   // Standard Color definitions
#include "textures.inc"   // Standard Texture definitions
//#include "functions.inc"  // internal functions usable in user defined functions

#declare f_mymarble=function{pattern{marble turbulence 2.575}}
#declare f_mygranite=function{pattern{granite turbulence 1}}

#declare LandContour=function{
 y+f_mymarble(x/100,y/100,z/100)*100
 +f_mygranite(x,y,z)/10
}

isosurface {
  function { LandContour(x,y,z) }          // function (can also contain declared
functions

  contained_by { sphere{0, 1e3} }  // container shape
  //threshold 0.0                     // optional threshold value for isosurface [0.0]
  accuracy 0.01                      // accuracy of calculation [0.001]
  max_gradient 1                      // maximum gradient the function can have [1.1]
  evaluate 1, 1.2, 0.95             // evaluate the maximum gradient
  //max_trace 1                       // maybe increase for use in CSG [1]
  //all_intersections                 // alternative to 'max_trace'
  open                              // remove visible container surface


  pigment{gradient y}
}


#declare Place1=<0,100-clock*90,-100+clock*100>;
#declare Place2=<-clock,100-clock*90,-99+clock*99>;

camera{
 location Place1
 look_at Place2
}

light_source{<-10000,10000,0> color rgb 1}

// Create an infinite sphere around scene and allow any pigment on it
sky_sphere { pigment {color rgb <0.7,0.7,1.0>} }

Post a reply to this message

It seems reasonably fast for what it does. Complex isosurfaces are often
not very fast.

You can speed it up by a factor of about 3 by fitting a tighter
contained_by shape, like

  contained_by { box{<-1e3,-111,-1e3><1e3,0,1e3>} }

If it's important for it to have a circular horizon you can intersect it
with a sphere or cylinder. If you're lucky you may well be able to get
away without having to use max_trace which would eat into the speed
improvement.

-- 
Mike Williams
Gentleman of Leisure

Post a reply to this message

Bryan Valencia wrote:
> // Persistence of Vision Ray Tracer Scene Description File
> // File: IsoDessert.pov
> // Vers: 3.5
> // Desc: Tests using isosurfaces for landscapes
> // Date: 2/15/2005
> // Auth: Bryan Valencia
> 

What is your definition of a year?

On my system it took 17minutes.


-- 
Maurice

http://get-me.to/Hendrix

-----BEGIN GEEK CODE BLOCK-----
Version: 3.12
GEd-s+:+aC++UL+PL+E-W++N++o--K-
wO---M-VPS+PEY+PGPt++++5--XRtv+
b+DI+D--Ge++h----y++++
------END GEEK CODE BLOCK------

Post a reply to this message

Actually it did the sky in a jiffy, then it did 1 line every 4 minutes at
the horizon, and sped up a little after that.  At the time I posted, I had
just waited for the first few lines of the horizon, and cancelled, assuming
the whole isosurface would be as slow.

Also, I am confused as to why the 'land' all seems to be below y=0 when the
code is adding functions to y.


"Maurice" <cel### [at] nospamhotmailcom> wrote in message
news:42122653$1@news.povray.org...
> Bryan Valencia wrote:
> > // Persistence of Vision Ray Tracer Scene Description File
> > // File: IsoDessert.pov
> > // Vers: 3.5
> > // Desc: Tests using isosurfaces for landscapes
> > // Date: 2/15/2005
> > // Auth: Bryan Valencia
> >
>
> What is your definition of a year?
>
> On my system it took 17minutes.
>
>
> -- 
> Maurice
>
> http://get-me.to/Hendrix
>
> -----BEGIN GEEK CODE BLOCK-----
> Version: 3.12
> GEd-s+:+aC++UL+PL+E-W++N++o--K-
> wO---M-VPS+PEY+PGPt++++5--XRtv+
> b+DI+D--Ge++h----y++++
> ------END GEEK CODE BLOCK------

Post a reply to this message

Out of curiosity, what settings did you render with?

With no AA and at a resolution of 800x600, it rendered on my P4 2Ghz
laptop in just under 4 minutes under Linux (self-compiled POV-Ray build
running at niceness -20).

Jim

Post a reply to this message

Wasn't it Bryan Valencia who wrote:

>Also, I am confused as to why the 'land' all seems to be below y=0 when the
>code is adding functions to y.

The surface exists at points where the whole function evaluates to zero.
The functions you added to y can take values between 0 and 100.1. So to
get the whole function to add up to zero, y must take values between 0
and -100.1.

#declare LandContour=function{
 y+f_mymarble(x/100,y/100,z/100)*100
 +f_mygranite(x,y,z)/10
}

E.g. suppose for some values of x, y and z
         f_mymarble(x/100,y/100,z/100) evaluates to 0.5
     and f_mygranite(x,y,z) evaluates to 0.5
then your LandContour surface exists where
         y + 50 + 0.05  =  0
So y must be -50.05

-- 
Mike Williams
Gentleman of Leisure

Post a reply to this message

My current version of this is on a 1GHz Pentium is taking 1H 3m.  The fun
part is that it's intended to be part of an animation.

Now I remember why I keep going back to heightfields.



"Jim Henderson" <nos### [at] nospamcom> wrote in message
news:pan### [at] nospamcom...
> Out of curiosity, what settings did you render with?
>
> With no AA and at a resolution of 800x600, it rendered on my P4 2Ghz
> laptop in just under 4 minutes under Linux (self-compiled POV-Ray build
> running at niceness -20).
>
> Jim

Post a reply to this message

I know the feeling - I've been playing a lot lately with radiosity, focal
blur, and photos, and my renders are taking anywhere from an hour to (the
longest so far) 65 hours on the 2 Ghz system.

But the results are quite stunning. :-)

Jim

Post a reply to this message