Well, here's some more tests.  This one was done with samples at 20 and took
about 36 min.  As you can see, there's no AA on this one, though.  Even at 20,
you can still see that "shelving".

I suppose one could cut up your atmosphere into segments, and have something
like:

intersection {
  sphere {0, AtmosphereSize}
  box {<-Size,0,-Size>, <Size, CloudAlt, Size>}

  ...samples 1,1...
  density {MediaFunction(x,y,z)}  //all using the same function
}

intersection {
  sphere {0, AtmosphereSize}
  box {<-Size,CloudAlt,-Size>, <Size, CloudAlt+CloudDepth, Size>}

  ...samples 20,20...
  density {MediaFunction(x,y,z)}
}

But I suspect that in some way the tangent surfaces would show up.  Also, I'm
finding that changing the samples tends to affect the apparent density.  So the
density of the background might be obviously different in the two boxes.

Post a reply to this message

Attachments:
Download 'lightinthevalley02.png' (383 KB)

Preview of image 'lightinthevalley02.png'