RE: 16 bit editing...

["Jon and Amy Harris" <jonandamy@foxinternet.com>]

When I saw the 8- vs. 16-bit images I thought the same thing--that the
contrast was significantly different between the two which wouldn't normally
be an artifact of 8-bit editing.  I also played around a little bit trying
to make the 8-bit closer to the 16-bit but you did a much better job.

So it still seems no one has yet posted a "real world" image that shows the
benefits of 8- vs. 16-bit editing made using typical adjustments.  It is
very easy to show differences with an artificial image and/or ridiculous
curves/levels adjustments, but of course that doesn't prove much except--and
this is my main point--that 16-bit editing is more forgiving to doing things
the "wrong" way.

If you made many tonal/color adjustments some which compress the tonal
range, and others that expand it, some significant posterization could
result.  But since you are making competing adjustments, this isn't the
"right" way to do it.  Similarly, if you scanned your image so that it
arrived using only a small part of the tonal range, and then tried to expand
it using 8-bit editing, that could also create posterization.  But again,
the "right" way to do it is to make adjustments in your scanning software so
that the resulting image fills (or nearly fills) the entire tonal range.  If
this is the case, then the adjustments you would need to make in 8-bit land
would be minimal and not very destructive.

In the cases above, 16-bit editing would bail you out and you could still
have an artifact-free image even when doing things totally wrong, so that
certainly is a benefit for some.  However, if you make sensible adjustments
and use adjustment layers for these then the benefits are probably not very
great compared with the cost of increased processing time and storage space.

Keep in mind that in 8-bit mode, you can still store 16,777,216 colors which
is supposedly more than the human eye can perceive.  Anyway, that's my take
on the 8- vs. 16-bit issue.

An aside, as an electrical engineer who works with digital audio, there is
an analogous situation with 16- vs. 24- vs. 32-bit audio recording and
editing.  Again, if you don't need to make significant level adjustments,
the higher bit editing doesn't help you much.  But using the higher bit
modes can be useful in that you don't need to be nearly as careful about
your levels--it is more forgiving.  In audio, the penalty isn't as bad--a
50% increase in storage vs. 100% in graphics--and there aren't restrictions
where you can't use certain tools in higher bit modes like PS has.  So the
tradeoff generally comes down much more favorably toward higher bit work.
But the issues are still the same.

-Jon

> -----Original Message-----
> From: owner-epson-inkjet@leben.com
> [mailto:owner-epson-inkjet@leben.com]On Behalf Of Todd Flashner
> Sent: Monday, May 27, 2002 9:02 AM
> To: epson-inkjet@leben.com
> Subject: Re: 16 bit editing...
>
>
> Good job Michael. That's closer to my expectations too.
>
> Todd
>
> >
> > Mark,
> >
> > Thanks for posting. That's the first time I've witnessed anybody posting
> > anything for examination when this topic comes up. I appreciate it.
> >
> > Now, in examining the images and their histograms, what I see
> is a lack of
> > contrast in the 8 bit edited image relative to the 16 bit
> edited in the lower
> > end. If the exact same curve adjustment operations were
> performed, I don't
> > understand how this difference came about. Anyway, when I
> adjust the 8 bit
> > image's curve lower end to closely match the 16 bit image
> curve's lower end,
> > then the results are almost identicle.
> >
> > I've posted the 8 and 16 bit edited images that you posted
> untouced. I've also
> > posted the adjusted (or corrected as I call it) 8 bit image.
> Alos, I've posted
> > the 8 and 16 bit histograms along with the 8 bit corrected
> histogram. If you
> > notice, the orginal 8 bit image has a relatively narrow range
> in the low end.
> > Narrow ranges will ALWAYS lead to dull/flat looking images. The
> 16 bit and 8
> > bit corrected ranges are pretty much the same and thus, the
> results are pretty
> > much the same.
> >
> > Navigate to http://greeraa.com/albums/bit_depth/ to see.
> >
> > --- mrkeene@attbi.com wrote:
> >> Mike,
> >>
> >> Here is a real world comparison for you.
> >>
> >> http://www.pbase.com/mark_kn/16bit_vs_8bit_editing
> >>
> >> The image was taken with a Canon D60 in RAW mode and converted
> to a 16 bit
> >> TIF.
> >> A duplicate was made and converted to 8 bit.
> >> An equal amount of levels were applied to both images.
> >> The files were then resized to 640x480 for speed of loading
> and saved as JPG
> >> quality 10.
> >> No other manipulation was performed.
> >>
> >> Download the images (they are ~80k each). Look at the
> original. Notice how
> >> far
> >> down in the shadows the lower portion of the image is. Now observe the
> >> Johnson
> >> grass in the lower right foreground on the 16 bit edited file
> compared with
> >> the
> >> 8 bit edited file. Draw you own conclusions.
> >>
> >> Regards,
> >>
> >> Mark K.
>
> -
> Turn off HTML mail features. Keep quoted material short. Use accurate
> subject lines. http://www.leben.com/lists for list instructions.

-
Turn off HTML mail features. Keep quoted material short. Use accurate
subject lines. http://www.leben.com/lists for list instructions.