Am 16.03.2017 um 17:36 schrieb Mike Horvath:

> This works a bit better, but is still not right:
> 
>     #local nSum = 0;
>     #for (i, 0, 89)
>         #local nSum = nSum + (xyz_table[i].y * D65_table[i]);
>     #end
>     #local xSum = 0;
>     #for (i, 0, 89)
>         #local xSum = xSum + xyz_table[i].x * 1 * D65_table[i];
>         #local ySum = 0;
>         #for (j, 0, 89)
>             #local ySum = ySum + xyz_table[j].y * 1 * D65_table[j];
>             #local zSum = 0;
>             #for (k, 0, 89)
>                 #local zSum = zSum + xyz_table[k].z * 1 * D65_table[k];
>                 #local xyYCoo = 1/nSum * <xSum, ySum, zSum>;
>                 #local xyYCoo = XYZ2xyY(xyYCoo);
>                 #local xyYCoo = <xyYCoo.x,xyYCoo.z,xyYCoo.y>;
>                 sphere
>                 {
>                     xyYCoo, 1/80
>                     pigment {color rgb x}
>                 }
>             #end
>         #end
>     #end

Sorry, but this is totally bonkers.

The indices into xyz_table are to be interpreted as representing
wavelengths.

So you're varying three wavelength variables independently: For one (i)
you're summing up the x component; for another one (j) you're summing up
the y component; and for yet another one (k) you're summing up the z
component.

What good do you expect this to do?

What you really need is a smart external loop or set of loops that
generates /spectra/, not wavelengths -- and then within that loop insert
a /single/ additional loop that varies a wavelength variable i, summing
up xyz_table[i] * D65_table[i] * spectral_power(i).

Post a reply to this message