QTR-Quadtone RIP

> I'm wondering if when applying different color inks, even different
> black inks, how the printer applies these. Are the colors "mixed" so
> that a particular dot of ink can be a mixture of more than one ink
> color? And to do this, what would the QTR curve settings (density,
> limit) be say if I wanted to lay down only red dots made up of magenta
> and yellow inks?   
> 
> Or does it lay down the individual colors as separate dots that the
> eye sees as a new color--for example, a red color is acutally made up
> of separate magenta and yellow dots of ink. 
> 
> Best, David.
>
Both.
Printing in essence is a binary process : a drop or no drop (let's put 
aside dot size for simplicity). On the paper, a color drop alone will 
render that color. If you lay down 2 drops, both will have their 
corresponding primary color.  Now when many drops are laid down some 
will be covering or overlapping each other so an M dot will partly 
cover a Y drop and their will be 4 fractional area coverages : white, 
Y, M and M+Y (R). This process has been well defined by Neugebauer (for 
litterature reference) and a CMJN model delivers 16 different 
fractional area coverages with their own spectral reflectance. Besides, 
light reflection is not just the light hits the surface then is either 
absorbed or reflected, a part is penetrating into the substrate because 
ink is translucid, scaterred and reflected back out of the paper at any 
probabilistic point (can be from where it entered or else), as such you 
have both a substractive and additive color model. The Neugebauer model 
doesn't not take that into account and is being modified by Yule 
Nielsen by an empirical factor (let's call it dot gain correction 
though it's not).
QTR can't do that because many factors are unknown. For instance, you 
can mesure the overall reflectance of a given sample but not the 
individual fractional area coverages to sum them up. However, you don't 
either really need it because QTR curve tool partitions inks based on 
relative density corrected of dot gain. Since many combinations of 
curves will output the targeted levels (BO would also for instance), 
all that is needed is to make the overall output linear. Now add one 
additional variable which is granularity (hope this is the right 
English word, correct me if wrong), and that becomes another game 
because there *is* then a single combination that delivers both best 
achievable linearity and smoothness...but be ready for frustration.

Olivier