[filmscanners] RE: 8 bit versus 16

["Austin Franklin" <austin@darkroom.com>]

Robert,

> Yes - 8 bit does work fine for most images, but if
> you really want to throw an image into some editing,
> then relying on 8 bits is foolhardy if you can get
> more to work with.

BUT...you really don't GET 16 bits.  You get 10, 12 or 13, and even if you
*think* you get 14, you really don't.  It also depends on if you are
scanning slides, which will use more bits, or negatives, which will end up
occupying less bits.  It's just how scanners work.  Scanners read relative
density, and that's it.

> Remember - filmscanners work with an analogue medium
> that contains far more information than 16/8 bits
> can capture

Than 8 bits yes, but it's no where near more than 16 bits.  You are lucky to
get 10 if even 11 bits out of negative film.  10 bits is a density range of
3.0, 11 bits is a density range of 3.3 and 12 bits is a density range of
3.6.  Have you ever measured the density range of color negative film?
Provia has a stated dMax (in the Kodak data sheet) of 3.0, and a dMin of .2,
which gives a density range of 2.8.  A density range of 2.8 requires only 10
bits.  And, keep in mind that because you have a density range of N, does
not mean that you actually have a FILM color resolution that will allow
discernability of all those bits!  This is particularly true of slide film.

> - now why not only use 4 bits? or 6/7?
> 8 Bits is no magic number...

Well, for grayscale, it's more than they human eye can discern (which is
around 100+), so that's why not 4 or 6.  Though, 7 would do fine for
printing an image.  But, because we can't discern tones, doesn't mean that
they aren't useful...because you want the tonal transitions (if they are
that way in the original scene that is) to appear smooth, and you can only
do that by using indiscernible tones...if they were discernable, you'd see
them ;-)

> - just as the 16 Million
> colours is a myth
>- in the sense that no digital
> image contains all 256*256*256 possibilities.

I don't see the "myth" you believe...no one ever said that ALL possibilities
are in any image.  You're missing the point.  What is important is what they
eye can discern, and that you can represent all the variants that an eye can
discern for any image (within the limits of the color space chosen that is).
It's the gradient that is important, not the overall number of colors.

There are also two different issues.  One is tonal curves, which is a
different issue than printability/viewability.  The second is based solely
on our ability to discern colors, and that is very well scientifically
documented what the limits of human vision are.  The first is based on how
much of a tonal movement can you do and it not be discernable in the output,
and that is entirely image, and amount of tonal movement, dependant.

> Heyy - it might contain 3400*120*44 ...
>
> Its well documented in the 3D community that having
> 24 bit colour internally in 3D processing engines
> can result in banding in certain scenes, and thats
> why Nvidia and ATI have developed 32 bit engines,
> and more.

That's an entirely different issue.

> I think that scanning to capture all the nuances
> and working from there is the sensible way.

I understand you (and some others) believe that, but that doesn't mean it's
true, as a general rule, or that there is any benefit from it, as a general
rule.

Regards,

Austin

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