QTR-Quadtone RIP

I am surely not a pro to advise but I had a lengthy discussion both 
with Roy and Steve Kale for several days and thanks to their good 
will I understood more or less the thing. I believe most of the talks 
are on the BWdigital group. Meanwhile some replies :

> Now however, using the densitometer readings, aside from getting a  
> smooth ramp, after linearizing and achieving near ideal densities 
(at  
> least as far as what is listed in the available documentation) the  
> images print quite dark when printed directly and require a PS  
> adjustment layer to print properly. It seems to me that the curve  
> needs to be lighter, but if I make it lighter then the densities 
will  
> no longer match the ideal values.

This is because the linearization just distributes densities evenly 
(in even steps): the process does not map source gray values to 
destination gray values as per an CM worklow. Consequently prints 
look dark and this is why Roy's worked extensively on the ICC 
droplets.
> 
 
> So, in the case of BW, I assumed that since we're working  
> monochromatically, there wouldn't really be any need for 
profiling,  
> as the linearization should result in the desired values. To be a  
> little more specific, since the process of linearization is 
setting  
> up the curves so that the printer produces specific densities when  
> receiving cmyk values in a single ramp (as opposed to mixing 
multiple  
> ramps for color mixing), the process of profiling the printer's  
> output would only be redundant as the only purpose would be to  
> profile the output of values since there isn't any color to 
consider  
> (aside from the issue of being able to soft proof with the correct  
> tonality).
> 
Precisely not. Printing with just linearization just produces a 
smooth ramp, but much darker, the values need remapping to produce a 
correct "lightness". 

> Since a color management system is meant to translate values as  
> accurately as possible from one device to another via the PCS, I  
> figured that what appeared to be 50% grey on my monitor would  
> correlate to the corresponding ideal density for the 50% patch on 
the  
> test print. But, this has not been the case each time I have 
redone  
> the process.

50% reults more or less in a 55-60 dark patch. If (in L* values) you 
get Dmax L20 and Dmin L95 you will see that the middle of it is about 
57-58 which is not mid-grey but which is linear.

> 
> So in specific, I'm hoping someone could clarify the following,
> 
> 1. If an ink/paper combination is properly linearized, should the  
> output be reasonably close to what's on the screen if the display 
has  
> been properly profiled? Or is it also necessary to profile the BW  
> output or use one of the other screen/print matching processes to 
get  
> this type of match?

You clearly need to process a BW profile that when visualise in a CMS 
shows a kind of gamma curve.
> 
> 2. What exactly would be the purpose of profiling BW output, aside  
> from being able to soft proof for tonal variations?

Making sure that source values are mapped to destination values.
> 
> 3. Since QTR will either take an image converted to 2.2 gamma and  
> print it without altering the values or in the case of images not  
> saved in 2.2 gamma, will convert to 2.2 gamma prior to printing, 
ie:  
> change the numbers to correlate to the previous color space,  what  
> would be the purpose of using one color space over another since 
what  
> is seen on screen and subsequently encoded in the 2.2 gamma space, 
is  
> what is printed?

QTR is not ICC compliant in the sense it does not care about the 
embedded profile. You have to convert to the BW profile, save the 
TIFF and then print the new TIFF. I saw you work on Mac, I assume the 
process would then be to concert to the profile and print direct from 
PS in this case.
> 
> 4. Are the ideal densities meant to correspond to our visual 
response  
> to changes in value? I thought this was the case but now I'm not 
sure.
> 
I'm not sure I understand. There's no ideal densities : what counts 
is to reproduce your screen on the paper (BPC exluded).

> 5. Although the curves are not profiles are they meant to be used 
as  
> such, or are they more like the Epson driver settings that just  
> control the ink limits etc. and should be used in conjunction with  
> destination profiles?
>

The curve linearises the output, the profile make sure L* are proerly 
converted source->destination.
 
> 6. Am I just way off base, regarding my understanding of this  
> process, or am I headed in the right direction?
> 

It personally took me about a month to (kind of) understand and I 
went thru the same questions (precisely!!!).

Take a look at Roy and Steve's explanation on BWdig group if the 
above is not explanaroty enough.

All I can assert is that the ICC droplets change totally the quality 
of the print and you get results above expectations.

Olivier