[filmscanners] Re: 8 bit versus 16

["Arthur Entlich" <artistic-1@shaw.ca>]

Robert Logan wrote:


>
>>For most images, there will be
>>no visible degradation in the image using only 8
>>bits/color.  If you haven't tried an experiment, and
>>are only speaking of "theory"
>
>
> Ive noticed it in practice severally. Notably in
> shots with some very variable lighting across
> landscapes which contain numerous greens. Playing
> with the curves often results in problems when all
> 256 of the greens in 8 bit are used up. And as
> you point out - with grayscale - luminance can
> get blown out with too few bits, so why not chrominance.
>
>

It's not quite as "black and white" as this implies.  There are 256
brightness values of any one green hue.  That means there are actually
many hundreds of thousands of greens (well millions in fact), some with
differing levels of red and blue in them (keep in mind that when working
with light, green and red make for a more yellowish color.  So every
pixel not only has one brightness level of green from 256 (including a
zero level or no green), but also potentially one brightness level of
red and blue.  The likelihood of a green field being made up of only one
hue of green (let's say all with zero red zero blue), is pretty much nil
(it would probably be unesthetic to look at).

The real pallet in the 8 bit image is 256 paint buckets of different
brightnesses of red, 256 of green and 256 buckets of blue, and each
color on the screen is made up of equal quantities of paint from any mix
of any one red bucket, any one green bucket and any one blue bucket,
(including, one bucket that is basically empty in each color (value 0)).

So, let's say you have a green brightness level 186.  That could have no
red and no blue, or it could have no red, and 1 brightness level blue,
or 1 brightness level red and no blue, or 1 brightness level of red and
blue... and on and on.  Obviously, at some point the color would no
longer be considered green.  186 green, 185 red and 185 blue would be a
a very nearly neutral gray, but it would be very slightly green.  If it
was 186 green, 186 red, and 186 blue would indeed be a neutral gray.

So, stating there are "only" 256 levels of green is not accurate.  There
are only 256 brightness levels of any one hue of green, but many
hundreds of thousands, in fact, millions of what most people would refer
to as greens.

The vast majority of human eyes out there cannot actually accurately
make the distinctions offered in a 16 million color pallet.

I question that you are actually able to see the difference, especially
on a computer screen.  I suggest a "double blind" test and try to choose.

Art

----------------------------------------------------------------------------------------
Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe 
filmscanners'
or 'unsubscribe filmscanners_digest' (as appropriate) in the message title or 
body