> The scene he posted didn't use any triangles, just a CSG difference with
> two superellipsoids. This is probably an error in the solving method,
> maybe some intersections are not being found.

I meant that... (some internal POV tesselation error).

> BTW, the shape of a "rounded box" is not exactly the same as a
> superellipsoid...the superellipsoid doesn't have any truely flat areas.
>
> A CSG shape might actually be faster than a similar superellipsoid...it
> will use more memory, of course. Another solution would be to use a
> difference of two isosurface superellipsoids...isosurfaces use a
> completely different solving method which might render without
> artifacts, and can sometimes be faster...though harder to use.

Actually RoundBox is very close on the flatish areas, and differs more
where it actually curves: the superellipsoid follows a parabolic. Here's
a scene to show the difference.

camera {
 location  <0, 0, -3.0>
 look_at   <0, 0,  0.0>
 orthographic
}
union {
  object {RoundBox(<-1,-1,-1>,<1,1,1>,.5) pigment {Blue}}
  superellipsoid{<.25,.25> pigment{Red} translate <0,0,2>}
  translate <1,-1,0>
  scale <4,4,.25>
  translate <-1,1,0>
}

Maybe a superellipsoid could be more closely approximated by scaling some
of the RoundBox components then rotating them.
--
light_source{<-9,9,-9>rgb x+y+z}text{ttf"timrom.ttf"    // timothyea
"Tim Attwood".1,0 pigment{rgb x+y}translate<-2.7,-.3,5>}// @worldnet.att.net
sky_sphere {pigment{gradient y color_map{[.45 rgb<.2,.3,.2>][0.5 rgb x+y+z]
[0.55 rgb x+y][0.6 rgb x+.5*y][.7 rgb 0]}scale 2 translate -1}}

Post a reply to this message