Digital BW, The Print

Peter writes:

> The spectral response of both the film and
> the filters used in CCD Bayes patterns overlap
> (as do the photopigments in the human eye) so
> I'm not sure what you mean by "in between".

I simplified the illustration.  The fact that they overlap does not change
the principle, or the constraints.

My example of the yellow-sensitive film illustrates this.

If RGB were sufficient to represent an entire spectrum from the original
scene, then none of the color systems used by human beings would work,
because they all depend on the inability of an RGB system to distinguish
certain spectral characteristics.  There is no way to distinguish between
monochromatic yellow and a mixture of red and green for example, but in real
life there _is_ a difference ... it's just that the information required to
detect that difference is missing from RGB.

> Are you saying there are wavelengths of light that
> fall between the response wavelengths of the color
> film (or CCD element filters) that the black and white
> picks up?

No, I'm saying that three numbers do not tell you which intensities at which
frequencies produced the RGB values.  Each number is a summation of many
different frequencies of light, and all distinction between them is lost.
But since you need to preserve this distinction in order to reconstruct the
curves you need to translate from one colorspace to another, the net result
is that RGB cannot be used to convert from one color or B&W film or sensor
to another.

> Could you give a specific example?

I already did.  I'll copy and paste it here:

You have a special film (or a special filtered CCD) that is sensitive only
to the yellow light of low-pressure sodium-vapor lamps.  You take a picture
of a scene that is lit with these lamps in some areas, and lit with a blend
of red and green light in other areas.  You then take the same picture using
an ordinary color film or color CCD.

What will you get?  On the special yellow-sensitive film, the areas lit by
LPS lamps will appear bright, and all other areas will appear dark.  On the
color film, however, all areas will appear equally bright.

There is _no way_ to convert the RGB image from the color film in a way that
will duplicate the results you got from the yellow-sensitive film, because
the important information on the spectral distribution of the original light
is gone.

> I don't know what you mean by the "curve of the original scene".

If you pass the light from a spot in the original scene through a prism,
you'll see the spectral distribution of the light from that scene.  That's
what you need in order to convert from one colorspace to another.  Just an
isolated red, green, or blue value won't do.

> All you need to care about is the curve(s) of
> the recording medium.

The curve of the recording medium won't do you any good unless you have the
curve of the original scene as well.

> I don't understand what information you think is
> missing.

The spectral distribution of the light in the original scene.

> You keep talking about data "in between" the R,G, and B
> values but those sensitivy curves overlap.

They overlap, but that isn't enough.  If that were enough, then there would
be virtually no such thing as color blindness (which typically involves only
a slight different in sensitivity curves), and no RGB color system would be
capable of producing the entire rainbow of colors for human vision.