Digital BW, The Print

> From: Tim Atherton [mailto:timatherton@...]
>
> > There are two parts to this. First of all, the author is recommending
> > conversion to B&W by desaturation. This is the minimal way to do the
> > conversion, and loses all the nice control that is available by
> using the
> > channel mixer, a grayscale proof setup, or the layers technique recently
> > mentioned.
>
> No he isn't - he's assuming it's already an B&W/greyscale image, which you
> already have as an RGB - I think the desaturation is in order to ensure
> there is actually no residual colour tint in the image. Not sure if it's
> actually needed though

Okay, that makes more sense.

> > What ought to work would be to go into the Advanced mode,
> > select Color Controls, set Gamma to 1.8, leave Color Mode on
> Standard, and
> > leave the sliders at zero. Of course, you could do the same thing
> > by telling
> > PS that the print space is Adobe RGB and selecting a gamma of 2.2 in the
> > driver.
>
> However, that doesn't work - or at least it certainly doesn't give as
> neutral a print as this - selecting Colourmatch as opposed to
> Adobe RGB does
> seem to give a much more neutral print. I'm guessing that even though the
> image is a B&W one, because it's in an RGB colour space, the Colourmatch
> space is reducing the gamut of those areas of the space which in Adobe RGB
> are rendered by the printer drive as a green or blue tint. Or
> something like that...

My guess is that it has more to do with ColorMatch RGB having a white point
of 5000K, as compared to 6600K for Adobe RGB. However, as long as the
rendering intent isn't set to Absolute Colorimetric, I would expect the CM
engine to cancel out this difference. Maybe I don't understand how the CM
engine deals with white points.

> > None of this is particularly unconventional. However, it
> assumes that the
> > printer driver, with Color Controls (if that's what he meant by
> Automatic)
> > selected, is accurately represented, at least for B&W, by a
> > simple gamma 1.8
> > (or 2.2) curve. Maybe it is. But the 2200 also comes with a set
> of actual
> > ICC profiles, for each standard paper/ink combo, that probably
> > describe the
> > printer even more accurately. So while there's no harm in testing
> > anything,
> > I'd guess that using a real profile (instead of just assuming
> > Colormatch RGB
> > or Adobe RGB) has a better chance of providing accurate results.
>
> Not to date it doesn't - none of them gives anywhere near a
> decent greyscale
> print. The Epson (and most other profiles) are garbage for greyscale
> printing. This definitely gives a print that's more neutral than fiddling
> around for hours with the Epson greyscale balancer. It's certainly not
> perfect, but without a RIP, it's about the best I've seen to date

I agree--which is why I bought Eye-One Photo. But if the Auto or CC setting
is better described by the ColorMatch RGB profile than the NCA setting is
described by the specific paper/ink profiles, that would indeed be a lucky
accident. Who knows--maybe it is.

I wonder how much printer-to-printer variation gets into the act. If all
printers that come off the assembly line really are identical, then there's
no excuse for Epson not providing better profiles.

--

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