I'm a commercial photographer and my wife is an exhibiting artist, so I
have insight into both worlds and
I've got 15 years plus of digital experience.
You don't mention what the originals are or what the end product is,
but our experience is
with fine art prints. We stay in 16-bit RGB for the whole process.
There is no advantage in grayscale
conversion other than economy of file sizes.
If an image is color and we want B&W then we use channel mixing. Color
correction can influence the tonalities, so we keep this as part of the
process. Predominantly using the green channel, since it has the least
noise as a rule, but the subject matter would dictate the conversion.
Allowing a channel to be more influential in the conversion is like
photographing B&W with a filter. The tonalities of the same color as
the channel that is allowed to dominate will appear lighter. Just like
B&W photography a yellow/green filter will render most subjects more
attractively - a mix of using more green channel and slightly more red
channel will make the prettiest B&W conversions. Better skin tone, more
definition in clouds, more varied tonalities in greenery etc. Unless
your original is already B&W, in which case the conversion will simply
strip out any errant color and it is still better to favor the green
channel.
A jet ink print from an RGB file will be converted to 7 channels
CcMmYKk (if using Ultrachrome Inks). Grayscale, 1channel, is just less
information for this conversion and most drivers and rips expect to
see RGB files. If the material is destined for the web; an 8bit RGB jpg
will be your final product, so still best to stay in RGB. I've heard
from those that know about what happens "under the hood" that every
operation loses data - think 1 bit - so I've concluded that it's best
to not do any interim mode conversions.
Hope this helps.
Ulf Skogsbergh
On Monday, July 18, 2005, at 10:09 PM, Will and Pam wrote:
This is not exactly a color management question but it does seem to be
in
the same universe anyway...
I am doing a black and white negative scanning project in which the raw
scans are 16 bit RGB (scanned as positives). They are then converted to
grayscale in Photoshop, and inverted. I've looked at various ways of
doing
this. Most recently I've been converting to Lab, selecting the L*
channel,
and then converting to grayscale. My armchair color management
knowledge
suggests to me that this is a good idea because it separates any color
information from luminosity, so that the resulting grayscale image is
truer
to the luminosity inherent in the original scan. It seems as if this
is a
more "accurate" way to do it. However, I do notice that in many cases
when I
select the L* channel, the entire image appears much brighter. I've
been
including a Stouffer step-wedge in the scans, and the lower densities
of the
step-wedge (and the similar densities in the image itself, especially
in the
range d.15-.60) really seem to change alot when the a and b channels
are
off.
If there is a question in there I suppose it would be:
In principle, is using the L*channel a good way to approach the
grayscale
conversion? Are there better ways I should consider? Is there a reason
why I
would be seeing such a shift in apparent luminosity when discarding
the a
and b channels?
