Digital BW, The Print

scrber writes:

> Sorry, I must be dumb, but I just don't get it.

Few people ever do, and I usually run out of energy before they show any
evidence of understanding.

> Why is shooting colour (RGB or whatever) worse than shooting
> B&W straight?

I'll try again:

RGB is worse because it collapses all the spectral energy of the
original scene into just three numbers.  There is no way to restore that
spectral distribution from those three numbers.  True black and white
capture (be it electronic or film) captures an image as a function of
the continuous spectral sensitivity of the capture medium and the
continuous spectral emission of the image being recorded.  There is no
way to simulate that continuous function with just three numbers.  So
there exists an infinity of black and white scenes that cannot be
accurately reproduced using a conversion from RGB.

> You mention loosing 2/3 data.  Don't you have 3 times more...?

See above.  RGB gives you three numbers per pixel.  The original scene
contained an infinite number of intensites across an interrupted
continuum of light frequencies.  The RGB capture is a one-way function
that distills this vast amount of spectral information into three
numbers.  It works fine if RGB is what you want as a final result; but
proper black and white requires the combined sensitivity curves and
image spectrum to which I allude above, and you cannot reproduce these
from just the three simple numbers of RGB.

For example, you can shoot a scene in true black and white with a
narrowband yellow filter, and you can shoot the same scene in color.  If
you try to reproduce the B&W results you got with the yellow filter
using any conversion from the RGB color image, you'll find that they
don't match; they may not even come close.  It can't be done, because
the color image is missing too much information.

> Or is it that the colour image is comprised of 3 (black and white)
> channels and that in the conversion process you can only take
> one of these (be it a blend of the three individual ones or not)
> as the final greyscale image.

Almost.  When you capture directly in black and white, your capture
device (electronic or film, it doesn't matter) records a single number
per pixel that is a function of _every visible frequency of light_
hitting that pixel.  When you convert from RGB to B&W, you get a single
number that is a function of just three other numbers.  There's no way
that the RGB conversion can come anywhere near the flexibility and depth
of true B&W capture.

> Thinking out loud, what if you took a picture of a largely b&W or
> desatureated scene in both RGB and with B&W film - do you have a 
> disadvantage in RGB then or not?

If the original scene is truly black and white--all frequencies of light
equally reflected by all points in the image--the true B&W capture will
yield exactly the same result as an RGB conversion.  But real-world
scenes like that are scarce.