Digital BW, The Print

Hi Tyler,

Ok, the light is definitely growing at the end of the tunnel here. I 
don't know which part of my brain was switched off when I was somehow 
assuming that 16-bit values could ever get to the driver, but of 
course under Windows (at least) there's no way to do this. The OS's 
bitmap structure can handles a maximum depth of 8 bits, end of story. 
Thanks to you Ernst for the digest of the past discussion on that 
subject, it's good to see that the later versions of IP5 do actually 
use the full 16 bits. Also, the discussion of dither in that thread 
sorta tallies with my previous hearsay about them trading smoother 
tonality in exchange for a larger dither cell.

Anyway, this realisation about the 8-bit limitation in the OS has 
changed my feelings about the non-linearity of the Epson driver. It's 
possible to get a smooth tonal scale with just 256 values, provided 
those values are reasonably evenly spaced in terms of perceived 
lightness (L*). Roy's approach to linearisation guarantees exactly 
that. If the driver response curve varies from this, there will be 
some larger gaps somewhere in the range, and it only takes the wrong 
image to make them obvious. Paul mentioned "microbanding" being an 
occasional problem with prints generated via the Epson driver, which 
I would strongly assume is a symptom of what I'm talking about. Paul 
if you're still following this thread, what d'you make of this? How 
do you define microbanding, as opposed to any other kind of banding?


A few small comments inline...

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "Tyler Boley" 
<tyler@t...> wrote:

<snip>
> 
> I think we're talking about the same thing, but just disagree a tad.
> I'll just wrap up my end of this by saying that the lengths Bill
> Atkinson had to go to to achieve his results border on the absurd, 
and
> by all reports he'll never do it again for any amount of money. Even
> then there are things I'm still not thrilled with about his 
profiles.

If the comments in my second para above hold water, then they will 
apply here too, so I can see why there might be unwanted visible 
effects even with his profiles.

<snip> 
> Also, mapping during profile conversions has little to do with 
Adobe,
> unless you are using the ACE engine. You do have several 
alternatives
> to Adobe's ACE, however the difference I've seen between them are 
very
> slight with ACE being as good or better than any of them.

I'm still in Windows-land, so ACE is best of a bad lot (AFAIA).

> Also, the
> profiling software itself is probably the most responsible for how
> colors map to the device space and here (I think) is really the 
larger
> problem. I've been hoping someone would have their algorithms refer 
to
> Lindbloom's perceptually uniform LAB space just to see what happens,
> but no one has stepped up to the plate as far as I know. The 
different
> packages I've used all seem to characterize a device just fine and
> similarly, but the decisions they make about LAB to device 
conversion
> remains an "art" they all do differently.
> I'm probably taking up too much of B&W list space with this color
> stuff, and I'm on the edge of spouting BS, I highly recommend the
> Colorsync list, many discussions to wade through, but much there to 
learn.

I'm headed there now. Just let me confirm we're talking about the 
same place; I'm assuming it's the one with the homepage at:

http://lists.apple.com/mailman/listinfo/colorsync-users

> 
> > Yeah, I hear that. I'm just trying my best to create a setup that
> > will give me results that are "perfect enough". As I said before, 
I'm
> > actually pretty satisfied with a good majority of the colour 
images
> > I'm getting from the Epson driver. Of course it would be nice to
> > increase the percentage even further =)
> 
> I've seen a few posts here and there about the current (official, 
not
> alpha or beta) Gimp drivers being more well behaved, and someone 
just
> asked on the Colorsync list about profiling over them, you might 
want
> to check it out.

Will do, thanks for all the help.

Mike