[filmscanners] Re: 8 bit versus 16

["Brad Davis" <DocBradd@mac.com>]

On 22/9/03 12:09, "Preston Earle" <PEarle@triad.rr.com> wrote:

> I've been thinking more about this 8-bit vs. 16-bit question, and one
> thing puzzles me and has generally been ignored in this discussion.
> Someone (Arthur, Austin, Laurie, ????) brought up the question of
> "noise" in image data, but that issue has been bypassed in these
> discussions in favor of other comments. Yet it seems noise is the reason
> high-bit data is superfluous. What I'm thinking:
>
> 1. High-bit data is very small compared to low bit data. The ninth-bit
> is only 0.25% of the value of the full tonal range of 0-100%.

Shouldn't this be 0.20% ? I think it is 1/500 (actually 512, but to a first
approximation - but maybe I miss something).  The issue is real though, the
smallest difference between two levels is 1/65,000.  I don't want to open
the noise issue again, but I do think it is unlikely that most scanners have
an intrinsic noise level on the order of 16 microvolts in any usual
environment.  I wonder if that is why my Polaroid 4000 scanner is rated as
only 14 bits.

>
> 2. All visible files are the product of a final resize/pixel-combination
> of some sort, at least until we get 2800x4200 or larger video screens.

I don't think this is relevant, there are places in this note where I
suspect that you may be confusing DPI and bit depth at a pixel.  This
conversation is only about bit depth.

>
> 3. When scanners measure and assign digital values to image elements,
> adjacent pixels are given discrete values that are generally different
> by more than 0.25%, that is, the precision of the measurement is less
> than 8-bits for adjacent pixels.

If using 8 bit scanning adjacent pixels may be the same or differ by at
least 1/256 - approximately 0.40%, assuming that the A/D is capable of such
discrimination (and it should be)

>
> 4. Image editing steps which spread existing pixel ranges over larger
> ranges do not create more precise intermediate values than the starting
> value's precision. If an intermediate pixel value must be created
> between two pixels whose values are 128 and 130 (8-bit), the value won't
> be more precise if the original values are 127.504 and 129.504 (16-bit).

Yes they do. Ignoring the issues of noise for a moment, the addition of 8
bit makes it possible for there to be 256 levels between 127 and 128.  The
number is no longer 128 (decimal) but 32,768 when the high value in a 16 bit
word.
>
> I don't know how typical CCD scanners scan at lower resolutions than
> their maximum. Whether by averaging pixel values along the CCD array and
> making larger steps along the film movement, or by some other way, they
> still end up with adjacent pixel values that differ by more than 1 unit.
> Knowing these values to .5-unit precision doesn't change the average
> values reported.

This is why I think you may be confusing DPI with bit depth, this comment
(above) refers to spatial resolution, not intensity resolution.  In this
issue, (8 bit vs. 16 bit) the same number of pixels exist in either case, in
an 8 bit file, they may differ by 1/256, no less, in a 16 bit file, they may
differ by as little as 1/65,536.  Thus if one reduces the range of values
(say in the levels setting in PS), there are still levels to provide
transitions in the latter case.  It is unlikely you will reduce the levels
to a point where it is possible to make a discrimination (thus the
appearance of banding).

The issue that has gone on for some time is that in B&W, there are only the
256 levels and an extreme correction (say to get a white in an underexposed
shot) may reduce the number of levels to a point where there is something
akin to posterization.  Austin has been arguing (I think) that in color
there are effectively 16,000,000 because most colors are a mixture of 3 - 8
bit channels and therefore one is unlikely to run out of shades.  Of course,
if a color has value in only one channel (say red), then the same problems
as with B&W may occur.
>
> For 16-bit processes to be relevant, wouldn't adjacent pixels have to be
> identical to more than 8-bit precision?

Yes, exactly, a 16 bit representation is more nearly continuous than 8 bit.
The fact that 8 bits look continuous is an artifact of our perceptual system
which can't discriminate that many levels ( and some of the work of printer
drivers to smooth things out for the resolution of the printer).  Most
people can't resolve much more than 2^5th (32) levels, some 2^6th (64), very
few much more than that.  ( Of course I presume everyone here can resolve
2^7th or 128 levels.)  Much of the variability in this measure is a function
of differing assumptions and different methodology in measuring it, the
generally accepted number is 6 +/-1 bits with few people getting much over 6
bits.
>
> Preston Earle
> PEarle@triad.rr.com
>
>
>
>
> ------------------------------------------------------------------------------
> ----------
> Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe
> filmscanners'
> or 'unsubscribe filmscanners_digest' (as appropriate) in the message title or
> body

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