Digital BW, The Print

In a message dated 12/12/2003 3:23:42 PM Eastern Standard Time, 
DigitalBlackandWhiteThePrint@yahoogroups.com writes:
Not true. With three channels of 8 bits, that's roughly equal to 9.5 bits,
because the maximum value you can get by adding three 255's together is 765,
which takes about 9.5 bits to represent.
Not true either. The numbers multiply, they don't add. 256 x 256 x 256 gives 
us 24 bits or 16.7+ million different values. Each bit double the numbers and 
can represent 1 stop.

Claude


[Non-text portions of this message have been removed]

> From: claudej1@... [mailto:claudej1@...]
>
> In a message dated 12/12/2003 3:23:42 PM Eastern Standard Time,
> DigitalBlackandWhiteThePrint@yahoogroups.com writes:
> Not true. With three channels of 8 bits, that's roughly equal to 9.5 bits,
> because the maximum value you can get by adding three 255's
> together is 765,
> which takes about 9.5 bits to represent.

> Not true either. The numbers multiply, they don't add. 256 x 256
> x 256 gives
> us 24 bits or 16.7+ million different values. Each bit double the
> numbers and can represent 1 stop.

No, THAT's not true. We're not talking about the number of possible
combinations of 24 bits, which is obviously over 16 million. We're talking
about the apparent resolution, meaning the number of apparent levels of
brightness you can perceive, or a colorimeter can measure. Believe me, with
three 8-bit color channels, equally weighted, the smallest change you can
see or measure when you change one bit isn't 1/16M of full scale, it's about
1/750 of full scale, because the brightness is roughly speaking based on the
sum of the three bytes, not their product. Since the gamma isn't generally
1, and since the three colors aren't perceived to be equally bright, the 750
figure isn't accurate, but it's more than 255 but probably averages out to
be somewhere around 750. It ain't 16M, or anything remotely like that.

--

Ciao,               Paul D. DeRocco
Paul                mailto:pderocco@...

the bottom line really is that you have plenty of data to create a 
good 8 or 16 bit _depth_ image from a rgb image file.  The fact is 
that you can combine them in such away as to yeild a huge number of 
different b&w reditions of a scene after you capture the image, a 
small subset of which will appear similar to commercially available 
b&w film... the others are left for the creative adventurer.

mark

PS  the channels do get mixed so black is 0.0 and white is 3x256 for 
a 24-bit color file going to b&w which then is scaled to your final 
image depth (ie. Paul is correct).

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "Paul D. 
DeRocco" <pderocco@i...> wrote:
> > From: claudej1@a... [mailto:claudej1@a...]
> >
> > In a message dated 12/12/2003 3:23:42 PM Eastern Standard Time,
> > DigitalBlackandWhiteThePrint@yahoogroups.com writes:
> > Not true. With three channels of 8 bits, that's roughly equal to 
9.5 bits,
> > because the maximum value you can get by adding three 255's
> > together is 765,
> > which takes about 9.5 bits to represent.
> 
> > Not true either. The numbers multiply, they don't add. 256 x 256
> > x 256 gives
> > us 24 bits or 16.7+ million different values. Each bit double the
> > numbers and can represent 1 stop.
> 
> No, THAT's not true. We're not talking about the number of possible
> combinations of 24 bits, which is obviously over 16 million. We're 
talking
> about the apparent resolution, meaning the number of apparent 
levels of
> brightness you can perceive, or a colorimeter can measure. Believe 
me, with
> three 8-bit color channels, equally weighted, the smallest change 
you can
> see or measure when you change one bit isn't 1/16M of full scale, 
it's about
> 1/750 of full scale, because the brightness is roughly speaking 
based on the
> sum of the three bytes, not their product. Since the gamma isn't 
generally
> 1, and since the three colors aren't perceived to be equally 
bright, the 750
> figure isn't accurate, but it's more than 255 but probably averages 
out to
> be somewhere around 750. It ain't 16M, or anything remotely like 
that.

> 
> --
> 
> Ciao,               Paul D. DeRocco
> Paul                mailto:pderocco@i...