> Hm, could you post the code for that? Just because I'm curious, my
> solution was different (and IMHO a little more intuitive):

I'm not claiming superiority over your solution.
The code (somewhat simplified, the simplifications untested) was

  // max_trace_level += 2

  camera {
    cylinder 3
    angle 360
    up -Length*y
    translate 2*Dist*y
  }

  union {
    cone {
      -Length*y  0,
      3*Length*y  4*Length
      inverse
    }
    cone {
      3*Length*y  0,
      -Length*y  4*Length
      inverse
    }
    pigment {rgb 0}
    finish {reflection 1}
  }

The object sits centered at the origin, is of length Length and y is the
axis of rotation. The camera sits on the same axis but shifted away from
the object. To see why this works, consider any plane that contains the
y axis. In terms of the camera this corresponds to a row in the image.
The conical mirrors form a rectangular diamond in this plane with a 45
degree angle both to the axis of rotation and to the direction of the
rays of vision. After two reflections, they head towards the object.

  Mark Weyer

Post a reply to this message