It's discussed ad inifitum and ad nauseum because a small number of
people claim there is no perceptable difference between 8-bit and 16-
bit images. This just misleads people.
Let's debunk the idea of trotting out a name and saying so-and-so
has debunked the argument. That's a fallacious argument. It's a
textbook example of the ipse dixit ("he says so") fallacy.
So, let's leave names out and deal with the logic of their
argument . . . Give us the logic why you feel we should expect to
see no difference, after applying nonlinear edits to the image.
Here's my real world experience . . .
If you take an image and do no editing to it, except to convert to 8-
bits and print or make a Web image, you'll likely notice no
difference at all.
If you do edit your images, conversion to 8-bits can quickly lead to
posterization that will be evident even on a low-end monitor and
with an inexpensive ink jet printer. All you have is 256 values for
each channel, and you only have to compress/expand some of the
values on one of the channels to get non-continuous tones that
generate visible banding. Think of skin tones or flower petals. They
can quickly show signs of banding that look like splotches.
You can even get visible banding with 16-bit images. It's less
likely, but not impossible with extreme nonlinear transformations of
the information can. Information is not uniformly distributed in the
typical image, and it can be skewed in many images. So you can
compress/expand tones so there are discrete breaks in the data large
enough to be visible in a print even with 16-bits per channel. With
8-bits, it's quite easy to do it.
I do agree with the comment that BO printing is much less likely to
result in banding. My comments addressed color images more than BO
images. However, few people do BO printing with inkjet printers.
Instead, neutrals are typically generated by combinations of CMY
with B and sometimes b and sometimes even cmy. Typically, neutrals
greater than midtone have no B or b whatsover, and neutrals less
than midtone have progressively more B or b as you move towards
shadows.
Grayscale images that are not BO rely on localized contrast
differences in the color image to map into contrast differences in
the grayscale image. Working to increase those contrast differences
can result in visible banding even after grayscale conversion.
Cheers,
Mitch
--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "J Michael
Sullivan" <michael@h...> wrote:
> There should be absolutely no perceptable differences with an
inkjet print no matter
> whether it came from an 8-bit or 16-bit file.
>
> This has been discussed ad infinitum on other boards. But don't
take my word on it:
> Dan Margulis' seminal book "Professional Photoshop" pretty much
debunks the need
> for 16-bit.
>
> J Michael Sullivan
>Message
Re: 16 Bit vs. 8 Bit for BO
2004-01-04 by Glenn Mitchell
Attachments
- No local attachments were found for this message.