Christian Froeschlin <chr### [at] chrfrde> wrote:
> Chaanakya wrote:
>
> > As opposed to the different colors being shown throughout the object, I am only
> > seeing the different 'shells' on one side of the object, which is quite weird.
>
> your problem is that your slab is not centered on the origin.
> Therefore, your "scale 1.00001" resolves the coincident surface problem
> for the inner surface in the opposite direction you're expecting.
>
> You might try something like this:
>
> #version 3.6;
> #include "colors.inc"
>
> global_settings {
>    assumed_gamma 1.0
>    ambient_light White
>    max_trace_level 256
> }
>
> camera {
>    location <0,1,3>
>    look_at <0,0,0>
> }
>
> light_source {
>    <100,100,0>
>    color White
> }
>
> #declare base = difference
> {
>    cylinder { <-1,0,0>,<1,0,0>,1.5 }
>    cylinder { <-1.00003,0,0>,<1.00003,0,0>,1 }
>    // center desired target part on origin
>    translate -1.25*z
> }
>
> #declare curved = union
> {
>    #declare I = 1;
>    #while (I < 5)
>      #declare D = 0.1*I;
>      intersection
>      {
>        object {base}
>        difference
>        {
>          box {<-D,-D,-0.5>,              <D,D,0.5>}
>          box {<-D+0.0999,-D+0.0999,-0.6>, <D-0.0999,D-0.0999,0.6>}
>        }
>        pigment {color <2*D,0,0>}
>      }
>      #declare I = I + 1;
>    #end
> }
>
> object {curved}
>
>
>
> Note that you can use cylinders instead of boxes if you
> don't actually want the corners.
>
> Also note that for ior the best approach might be to stack solids
> without differencing away the insides. As long as each solid is
> fully contained within the outer solid all light rays should
> traverse the correct ior boundaries.

You, my friend, are awesome :D  Thank you so much :)  That's exactly what I
wanted!  This thread is now closed :)

Thanks again!

- Chaanakya

Post a reply to this message

"Chaanakya" <nomail@nomail> wrote:
> Christian Froeschlin <chr### [at] chrfrde> wrote:
> > Chaanakya wrote:
> >
> > > As opposed to the different colors being shown throughout the object, I am only
> > > seeing the different 'shells' on one side of the object, which is quite weird.
> >
> > your problem is that your slab is not centered on the origin.
> > Therefore, your "scale 1.00001" resolves the coincident surface problem
> > for the inner surface in the opposite direction you're expecting.
> >
> > You might try something like this:
> >
> > #version 3.6;
> > #include "colors.inc"
> >
> > global_settings {
> >    assumed_gamma 1.0
> >    ambient_light White
> >    max_trace_level 256
> > }
> >
> > camera {
> >    location <0,1,3>
> >    look_at <0,0,0>
> > }
> >
> > light_source {
> >    <100,100,0>
> >    color White
> > }
> >
> > #declare base = difference
> > {
> >    cylinder { <-1,0,0>,<1,0,0>,1.5 }
> >    cylinder { <-1.00003,0,0>,<1.00003,0,0>,1 }
> >    // center desired target part on origin
> >    translate -1.25*z
> > }
> >
> > #declare curved = union
> > {
> >    #declare I = 1;
> >    #while (I < 5)
> >      #declare D = 0.1*I;
> >      intersection
> >      {
> >        object {base}
> >        difference
> >        {
> >          box {<-D,-D,-0.5>,              <D,D,0.5>}
> >          box {<-D+0.0999,-D+0.0999,-0.6>, <D-0.0999,D-0.0999,0.6>}
> >        }
> >        pigment {color <2*D,0,0>}
> >      }
> >      #declare I = I + 1;
> >    #end
> > }
> >
> > object {curved}
> >
> >
> >
> > Note that you can use cylinders instead of boxes if you
> > don't actually want the corners.
> >
> > Also note that for ior the best approach might be to stack solids
> > without differencing away the insides. As long as each solid is
> > fully contained within the outer solid all light rays should
> > traverse the correct ior boundaries.
>
> You, my friend, are awesome :D  Thank you so much :)  That's exactly what I
> wanted!  This thread is now closed :)
>
> Thanks again!
>
> - Chaanakya

Sorry, but there is one thing...for some reason, I seem to see slits or holes of
some sort when I render what you gave me.  Is this some inaccuracy on POVRay's
part?  If so, how can I fix it?

Thanks in advance!

- Chaanakya

Post a reply to this message

Chaanakya wrote:

> Sorry, but there is one thing...for some reason, I seem to see slits or holes of
> some sort when I render what you gave me.  Is this some inaccuracy on POVRay's
> part?  If so, how can I fix it?

Is this already when using ior? There is small gap between
each part to avoid coincident surface, which may cause extra
reflections that look unexpected.

Post a reply to this message

Christian Froeschlin <chr### [at] chrfrde> wrote:
> Chaanakya wrote:
>
> > Sorry, but there is one thing...for some reason, I seem to see slits or holes of
> > some sort when I render what you gave me.  Is this some inaccuracy on POVRay's
> > part?  If so, how can I fix it?
>
> Is this already when using ior? There is small gap between
> each part to avoid coincident surface, which may cause extra
> reflections that look unexpected.

No, it's when I use almost the exact same code you gave me.  All I did was
declare a new variable N (number of shells) and set it equal to 50.  If I set it
to 10, it's perfectly fine.

Post a reply to this message

Am 31.08.2012 17:27, schrieb Chaanakya:
> Christian Froeschlin <chr### [at] chrfrde> wrote:
>> Chaanakya wrote:
>>
>>> Sorry, but there is one thing...for some reason, I seem to see slits or holes of
>>> some sort when I render what you gave me.  Is this some inaccuracy on POVRay's
>>> part?  If so, how can I fix it?
>>
>> Is this already when using ior? There is small gap between
>> each part to avoid coincident surface, which may cause extra
>> reflections that look unexpected.
>
> No, it's when I use almost the exact same code you gave me.  All I did was
> declare a new variable N (number of shells) and set it equal to 50.  If I set it
> to 10, it's perfectly fine.

what's your max_trace_level setting?

Post a reply to this message

Chaanakya wrote:

> No, it's when I use almost the exact same code you gave me.  All I did was
> declare a new variable N (number of shells) and set it equal to 50.  If I set it
> to 10, it's perfectly fine.

I tried with 3.6 and 3.7 using I < 50 (and D = 0.02*I) but did not see
any problem. Please post the exact scene and render settings used.

Post a reply to this message

Christian Froeschlin <chr### [at] chrfrde> wrote:
> Chaanakya wrote:
>
> > No, it's when I use almost the exact same code you gave me.  All I did was
> > declare a new variable N (number of shells) and set it equal to 50.  If I set it
> > to 10, it's perfectly fine.
>
> I tried with 3.6 and 3.7 using I < 50 (and D = 0.02*I) but did not see
> any problem. Please post the exact scene and render settings used.

I think the problem is there only if I do not use antialiasing.  Anyway, here's
the code:

#version 3.6;
#include "colors.inc"

global_settings {
   assumed_gamma 1.0
   ambient_light White
   max_trace_level 256
}

camera {
   location <0,0,4>
   look_at <0,0,0>
}

light_source {
   <100,100,0>
   color White
}

plane {
  y, -10
  pigment { hexagon Green, White, Blue }
}

// box {
//   -0.5, 0.5
//   pigment { checker Red, Blue }
// }

#declare base = difference
{
   cylinder { <-1,0,0>,<1,0,0>,1.00003 }
   cylinder { <-1.00003,0,0>,<1.00003,0,0>,1 }
   // center desired target part on origin
   translate -1.25*z
}

#declare curved = union
{
  #declare I = 1;
  #declare N = 10;
  #declare my_ior = -2.3;
  #declare my_ior_increment = (-1 - -2.3)/(N);
  #while (I < N)
    #declare D = (1/N)*I;
    intersection
    {
      object {base}
      difference
      {
 box {<-D,-D,-0.5>, <D,D,0.5>}
 box {<-D+0.0999,-D+0.0999,-0.6>, <D-0.0999,D-0.0999,0.6>}
      }
      pigment { color <2*D,0,0> }
      // pigment { transmit 1 }
      // finish {
      //  ambient 0
      //  diffuse 0
      //  reflection {
      //    1
      //    fresnel on
      //  }
      //  specular 0
      //  roughness 0.001
      // }
      // interior {
      //  ior my_ior
      // }
      rotate <0,0,90>
      translate <0,0,2>
    }
    #declare I = I + 1;
    #declare my_ior = my_ior + my_ior_increment;
  #end
}

object {curved}

Without antialiasing, this produces random little 'holes'.  However, I think
that with antialiasing, it should be fine.

Thanks again for all your help!

- Chaanakya

Post a reply to this message

> Christian Froeschlin <chr### [at] chrfrde> wrote:
>> Chaanakya wrote:
>>
>>> No, it's when I use almost the exact same code you gave me.  All I did was
>>> declare a new variable N (number of shells) and set it equal to 50.  If I set it
>>> to 10, it's perfectly fine.
>>
>> I tried with 3.6 and 3.7 using I < 50 (and D = 0.02*I) but did not see
>> any problem. Please post the exact scene and render settings used.
>
> I think the problem is there only if I do not use antialiasing.  Anyway, here's
> the code:
> ...

> Without antialiasing, this produces random little 'holes'.  However, I think
> that with antialiasing, it should be fine.
>
> Thanks again for all your help!
>
> - Chaanakya
>
>
When you use antialiasing, you take additional sub-samples in some 
pixels and average them.
When you have any coincident surfaces, it cause lots of, semi-random, 
noise. That noise trigger the antialiasing. The averaging of that noise 
result in smaller dark pixels having a LOT less contrast. This 
effectively hides most of the problem.


Alain

Post a reply to this message

Chaanakya wrote:

> I think the problem is there only if I do not use antialiasing.

the gaps are a little big for this scale, replace the values
of "0.0999" with "0.9999" or "0.99999".

Post a reply to this message

Christian Froeschlin <chr### [at] chrfrde> wrote:
> Chaanakya wrote:
>
> > I think the problem is there only if I do not use antialiasing.
>
> the gaps are a little big for this scale, replace the values
> of "0.0999" with "0.9999" or "0.99999".

Thanks Christian!  That fixed it :)

Thank you to all of you who helped me resolve this problem! :)

- Chaanakya

Post a reply to this message