Christian Walther <cwa### [at] gmxch> wrote:

> What you're seeing on the right side in the second scene is the back
> surface of the sphere (through the transparent front surface). Its
> normal points inward, and therefore the slope pattern value is between 0
> and 0.5. Try replacing your sphere { 0, 1 } by a difference { sphere {
> 0, 1 } plane { -z, 0 } }.
>
>   -Christian

Thanks.  I have to confess, in my wacky theorizing I never thought of the
normals on the INSIDE surface being used.  But it makes perfect sense.
Right now, I can't for the life of me think of a use for that particular
"feature"...but give me time!!!

And it didn't occur to me to simply "wipe out" the offending part of the
sphere. A perfect solution. HOWEVER... if I change the index values of my
color map to, say,

[0.0 color rgb <1,0,0>]
[0.7 color rgb <1,0,0>]
[0.8 color rgbt 1]
[1.0 color rgbt 1]

then the offending part can't be truncated--because along with the normals
on the OUTER surface applying red to the sphere about 70% of the way from
left to right, the INNER normals are covering the surface 70% of the way
from right to left!  So it looks like a solid sphere, no transparency
anywhere.

I wonder if this was forseen when the slope pattern was conceived or
written. A solution, off the top of my head, would be to eliminate the use
of normals on the inside of an object. Though that may leave others WANTING
that feature; or may have other unintended consequences.

Ken




Ken

Post a reply to this message