Digital BW, The Print

It's mostly a matter of semantics.

But I think there are a few absolutes.  To me a stepwedge labeled 100, 95, 90...
better have the those numbers in the file.  If your 90 happens to have 94 in it
you not longer have a stepwedge -- you ought not to sample values there.
Your post mentioned various conversions of stepwedges -- this destroys the
stepwedge  --  it's NOT a stepwedge.  I can't emphasize that enough.

The important thing is that the K values are the actual numbers in the file, the
L values are the meaning of the numbers given a specific profile.  A stepwedge
ALWAYS has the same numbers,  ASSIGNing different profiles changes just the
meaning and therefore what you see as L values.

Your examples of converting files and find the same L values is just the definition
of what Convert does -- it finds the new file number that will lead to the same
L value.

-----------------

On a semi related issue, I looked at you PDF on Color Management.  You have 
basically a comparision of ABW vs QTR vs ICC.  The major point that is not 
mentioned is -- what's the profile attached to the stepwedge?  For the non-CM
cases this is irrelevant but for the CM cases its crucial to the results.  The fact
that a CM graph is different than a non-CM graph is natural since the goal is
very different.  And you'd get a different CM graph for different grayspaces.

One way to look at this is that a non-CM stepwedge print is printing evenly
spaced K values, but a CM print is printing L values that are spaced based on
the grayspace.  When you print a GG 2.2 wedge with CM what you are really
telling the driver to do is print a wedge with L values 0,1,5,13,20,26, ...
(these come from sampling L on a GG 2.2 wedge).

On the surface this may seem a picky technicality but it really is very different.

Roy

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, Steve Kale <stevekale@b...> 
wrote:
> 
> > From: Roy Harrington <roy@h...>
> > Reply-To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
> > Date: Thu, 20 Oct 2005 23:07:46 -0000
> > To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
> > Subject: Re: [Digital BW] ICC v. Transfer Function in Epson driver
> > 
> > Steve,
> > 
> > I think you kind of have this backwards.
> 
> No I think we are just talking across purposes.
> 
> 
> >All your Converts are changing the
> > file values.  
> 
> Correct
> 
> >By Converting you no longer have an even stepwedge.
> 
> Understood but then you are changing observations.  My point was that the
> visual separation of an image doesn't change.  If I work up an image in Gray
> Lab or GG2.2 or GG 1.8 and print it with conversion to the print profile
> each image will look the same.  So visual separation evident in the image is
> NOT dependent on workspace (unlike a Same as Source workflow).
> 
> Of course the colour represented by 90% grey in GG 2.2 is not the same
> colour as 90% grey in GG 1.8 - that's what is meant by the fact that these
> two colour spaces have different gamma.  90% grey in GG 2.2 is the same
> colour as 94% grey in GG 1.8.  (It is L* 6 in both cases because it is the
> same colour in both cases.)  90% grey in GG 1.8 is a lighter colour than 90%
> grey in GG 2.2 - and it will have a different L*.  If you are looking at two
> colours in an image (forget their corresponding workspace-dependent file
> values for a moment) their separation in print is not altered by workspace
> in a colour managed workflow.  So when thinking about what happens to your
> image on conversion to a print profile don't be confused about these gamma
> differences in workspaces.  In each case the file value sent to the printer
> is adjusted by the appropriate amount to render the proper colour.
> 
> This is what I meant by an illusion of sample.  If we choose to sample 94%
> grey then yes this colour and its "separation" from 100% K moves with
> workspace - or more accurately with gamma.  This IS the definition of gamma.
> But by so doing we are sampling different colours.  94% only has meaning
> when attached to a profile.  When we print an image we do not see a bunch of
> numbers on the page but a bunch of colours.  The separation of the print
> space is determined by the print space NOT the workspace.
> 
> (To emphasise the point, get a step wedge and tag it first with GG 2.2 and
> print it with the printer profile, then convert it to 1.8 and print it with
> the printer profile.  The prints will look the same regardless of your
> workspace.  In each case you are printing different file values and
> different observation points on each greyscale but satisfy yourself that you
> are printing the same colours.)
> 
> It's pedantic point but an important one if you are coming from a Same as
> Source workflow where file values are printed without conversion with
> respect to the print space.  In that case workspace does matter.  A lot of
> people get confused by this stuff.
> 
> >What you 
> > see
> > with the K and Lab values are the actual values.  K is calculated by
> > (255-gray)/2.55
> > and L is the actual calculated Luminosity based on the grayscale profile.
> > 
> > BTW, if you are looking at RGB values throughout this exercise they will
> > absolutely
> > confuse the situation to no end.  You would be seeing another profile
> > conversion to
> > the working RGB space.  The numbers are fairly removed from what's in the
> > data.
> > 
> > The one feature of PS that shows what's really there is the Histogram.  A
> > stepwedge
> > should have nice even combs there.
> > 
> > You need to Assign profiles to try the different grayspaces.
> 
> I would constrain this to:  you need to Assign a profile to see what colour
> the same file number represents in the new space.
> 
> Assign keeps the numbers and changes the colour.
> 
> Convert changes the numbers and keeps the colour.
> 
> When we send a target to a printer for profiling we do so Without colour
> management.  This is because we want to see how the printer reacts to
> getting a particular number (which is of course all they react to other than
> a good thump when they are clogged!).  We don't want this number changed at
> any time by the colour management module.  Collect enough reactions to
> numbers pairs and we can create a picture of how it will likely react to all
> possible numbers.  With this profile, the colour management module does the
> conversion of numbers sent (ie changes the numbers) so that the new numbers
> represent the same colours on paper - subject of course to its policies with
> respect to managing out-of-gamut colours.
> 
> Steve
>