Digital BW, The Print

> From: Roy Harrington <roy@...>



>> Are you sure Roy? Sure the data stays the same, but our view of it
>> changes (especially in the darkest tones) and we edit the image
>> differently to compensate for the new LAB working space.
>> 
>> What I don't understand, is that if I have a QTR curve that works for
>> gamma 2.2 space, then I would need to create a new curve for a LAB
>> working space. If I create such a new QTR curve that works for LAB
>> space and not gamma 2.2 then I would want to "convert to profile" when
>> changing from gamma 2.2 image to a LAB greyspace image.
>> 
>> If I don't create a new QTR curve, then I loose my visual screen to
>> print match, even if I can make the the original print while working
>> from LAB space. It's hard to write this stuff and have it make sense!
>> But it makes perfect sense to me <g>
> 
> I'd first like to say that the best thing to do is try things out and see.
> 
> But here is my reasoning.   The QTR curves/profiles aren't based on the
> photoshop gray space at all.  When you print and select "Same As Source"
> you are just passing the data through without modification.  QTR gets
> the raw data.  When you did all the linearization it was based on raw
> pixel data -- it was not at all dependent on the gray space.

Yes this is correct.  We measure the density printed for a the image file's
"raw data" and then feed this back into the equation via LINEARIZE=

>So the
> existing QTR profiles should work just fine as long as the pixel values
> don't change, hence Assign-Profile.

This would work if the print space was LAB but it isn't.

>Another way to look at it is: do
> I want to preserve the way the screen looks (use Convert) or the way the
> print looks (use Assign).
> 

This is correct.  But for most people, what they have on screen is what they
want to print and hence convert is the right answer.

Bruce,

Unless I am mistaken, and I am happy to be proven incorrect, when we tag a
file initially with a profile, eg Gray Gamma 2.2, we are giving reference to
the individual pixel values.  So when the file has a pixel with value, in 8
bit, of 128 it can be described as 128/Gray Gamma 2.2. 128/Gray Gamma 2.2
has a density of 0.66.  If our screen's profile were LAB - which of course
it isn't as it has its own profile! but just for illustrative purposes -
colorsync would take that 128/Gray Gamma 2.2 pixel and deliver the screen
not 128 but 138.  This is because 138 is L=54 (there is some rounding here)
or the grey we want to see.

Now let's send the image file pixel value to a printer calibrated such that
it matched LAB.  We send the raw pixel values as noted above. In our
example, let's look at 3 scenarios.  Firstly let's send the Gray Gamma file.
The printer gets sent value 128 and prints density of 0.73 (ie the density
equivalent of 128/256x100=50).  If we Assign LAB to the gray gamma 2.2 image
there is no translation of raw pixel values.  Instead the existing raw pixel
values take on their new meaning in the new space.  This is why when you
assign a profile its look changes on screen. The printer still gets 128 and
prints the same density but this time it matches the screen because the
image on screen has changed.   Now the last scenario.  We Convert and all
pixels with value 128 now have file values of 138, ie the pixel vales have
been remapped to their correct respective places in the new space.  The
printer prints L=54 and we get the grey we wanted.

Cheers

Steve