Digital BW, The Print

> From: ginnylady33
>
>   I've been scanning via my 9000 at 16 bits but ultimately burning
> files onto Gold CDs for storage at 8 bit. It saves much space. I'm
> wondering if this is an error in judgment.
>  I know from experience that, at times, something that you'd expect to
> be a 'better methodology' may not be. For example, I've done detailed
> testing with my 9000 scanning B&W negs at 1x/16 bit vs 16x/16 bit
> superfine mode. (Which takes forever to scan)

You should never store a raw unedited image from any source in 8-bit mode.
Once you've carefully compensated for the exposure, and worked on the
picture until you're sure you've brought out all the important detail, and
are ready to print, then and only then can you safely cut back to 8-bit.

Be aware, though, that the degradation that occurs with the lower bit depth
won't appear in all images, and may only cause a problem occasionally.
Rather than try to judge each image, it's simpler just to keep everything in
16-bit for archiving purposes.

Another thing that may be confusing things for you is that noise in the
image actually compensates to some extent for the shallower bit depth. If
you create an artificial noiseless image (or even use a noise filter on a
real image) you can create a situation in which, after editing, you see
posterization in an image. However, the presence of significant noise in an
image prevents this from happening. In effect, the noise allows the
representation of finer gradations than the simple number of bits would
suggest, through the mathematical process known as dithering. (This is how
inkjet printers work, where smooth gradations are represented with only one
bit of resolution: on or off.)

In other words, you may be able to mask the problem by using 1x scanning,
since that produces more noise. If you do 16x multi-scanning, to average out
some of the noise, you'll generally get better images, but you'll be more
susceptible to posterization if you cut back to 8 bits.

--

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