Am 17.09.2010 17:33, schrieb gregjohn:
> Thanks, but I know how to get a PNG image in front of the camera, and I see this
> as the main benefit of screen.inc.
>
> What I want is to place semi-transparent image in front of the camera, such that
> what I see is generally X% my PNG, 100-X% the regular camera.  I came up with a
> very inelegant way to do this, but it was so clunky and sensitive to the room
> lighting.
>
> I am wondering what would be the way to **texture** a box so it is a
> semi-transparent overlay on top of my scene.


     #declare P=0.6;
     plane { z, 0
       clipped_by { box { <0,0,-0.1>, <1,1,0.1> } }

       texture {
         pigment { image_map { png "foo.png" once filter all 1-P } }
         #if (version < 3.7)
           finish { ambient 1 diffuse 0 specular 0 phong 0 }
         #else
           finish { ambient 0 emission 1 diffuse 0 specular 0 phong 0 }
         #end
       }
       no_reflection
       no_shadow
       no_radiosity
       no_photons

       transform { ... }
     }

This gives a neat semi-transparent square image stretching from <0,0,0> 
to <1,1,0>, which isn't affected by scene lighting in any way, nor does 
it affect the scene in any way - ready to be translated, scaled, 
rotated, sheared or otherwise transformed to be positioned right in 
front of the camera. P=0.0 gives you only the scene, P=1.0 gives you 
only the image, P=0.5 gives you an 50%:50% blend.

Depending on what effect you're /really/ aiming for, you may want to use 
yet some different voodoo though. For instance, you might actually want 
to "add" the image rather than blend it with the scene. For that purpose 
you could use something like:

   #declare P=0.6;
   #declare E=0.001;
   texture {
     pigment { image_map { png "foo.png" once filter all 1-E } }
     #if (version < 3.7)
       finish { ambient P/E diffuse 0 specular 0 phong 0 }
     #else
       finish { ambient 0 emission P/E diffuse 0 specular 0 phong 0 }
     #end
   }

Here, P=0.0 gives you the scene + 0% of the image, while P=1.0 gives you 
the scene + 100% of the image. E is just a very small value (what 
mathematicians would call "epsilon") that governs how well the whole 
construction really matches an "add" operation - the smaller the better 
(note that E=0.0 unfortunately doesn't work).

Post a reply to this message