On 05/16/2016 12:22 PM, EdR wrote:
> Hi,
>
> I'm having a problem rendering emitting objects, where transparent objects seem
> to be blocking the light from others behind.
>
> It can be seen in the minimal code (part of a larger scene) below and in this
> render https://www.dropbox.com/s/fki0av19wjwnbvu/minimal_bug_example.png?dl=0
>
> The left hand cylinder seems to obscure the right hand one, and I can't work out
> why. Can anyone help?
>
> Thanks!
>
> ----
> #include "colors.inc"
> camera {
>    location <15,0,40>
>    look_at <5,0,0>
>    angle 45
> }
>
> //background
> background { color Black }
>
> #declare thick = 1.5;
> union {
>    #for (ii, 8, 9)
>      cylinder{
>        (ii*thick*x), (((ii+1)*thick*x)-0.000001), 1
>        pigment { Clear }
>        hollow
>
>        interior {
>          media {
>            emission White * 0.03
>          }
>        }
>      scale <1,12.4,12.4>
>      }
>    #end
> }
> ----
>
>
>
>

I believe you are running into an internal a numerical cut off in the 
media sampling code(1). If you increase the overlap between cylinders, 
things work in my tests.

In addition, it would be more normal to use a merge (which eliminates 
internal boundaries) when using media or transparency vs a union - but 
maybe you want some different effect. The pigment and hollow can also be 
applied to the entire merge/union instead of each cylinder when they are 
identical.

My version of your code following some of the advice above :

   merge {
   //cylinder{ (1.5*x), ((3.0*x)-0.000001), 1 pigment { Red } }
   //cylinder{ (3.0*x), ((4.5*x)-0.000001), 1 pigment { Green } }
   //cylinder{ (1.5*x), ((3.0*x)-0.00001), 1 pigment { Clear } } // Fail
   //cylinder{ (3.0*x), ((4.5*x)-0.00001), 1 pigment { Clear } }
     cylinder{ (1.5*x), ((3.0*x)-0.0001), 1 pigment { Clear } } // OK
     cylinder{ (3.0*x), ((4.5*x)-0.0001), 1 pigment { Clear } }
     hollow

     interior {
       media {
         emission White * 0.03
       }
     }
   //scale <1,12.4,12.4>
   }

Bill P.

(1) - How individual rays move through the shapes likely will also 
affect where this accuracy / cut off issue is seen. As would, I expect, 
changing the media sampling criteria.

Post a reply to this message