Digital BW, The Print

"slcphoto73" <slchapin41@...> wrote:
>
>... - what do you mean by "linearize"? - in a simple way. 
> 

See the graph on page 11 of http://www.paulroark.com/BW-Info/Eb1400.pdf

We use 21-step (usually) test files to print test strips and then measure the Lab L or density values of the test strips.  The test patches of the test file are evenly spaced.  We'd like the values of the test strip to also be even or coincide with the distribution that is being displayed on our monitors.  Before "linearization" of a profile -- the last step in making one -- the output may be such that the values are not evenly spaced.

You can see from the graphs on page 11 that QTR uses a literally linear -- straight line -- relationship between the input values and the output (print Lab L) values.  

When making a QTR profile, the last step is to print a 21-step test strip with the pre-linearized profile.  The print from that is read with a spectrophotometer (or flatbed scanner using a calibration system) and the results of that measurement are fed back into QTR.  QTR then adjusts the profile such that the output from the linearized profile will be the straight line shown on page 11.

We usually use working spaces that are not characterized by literally linear relationships between the file values and monitor view or print values.  Gray Gamma 2.2 is the gray working space that is a subset of Adobe RGB, and is probably the most popular working space.  You can see from the graph that it is characterized by compressed shadows.

Because we all use color managed systems (which are best when calibrated), we're always in some working space, whether we know it or not.  Even if the file is "un-tagged" the view on the monitor is "color managed" and is displaying an image that reflects some working space input-output relationship.  

We'd like our prints to match what we see on the monitor.  "Linearization" is commonly used to describe the step we take to achieve this even if the final relationship is not technically linear.

For example, when we use QTR's "Create ICC" to make ICCs our "linearization" step does not result in a literally linear output.  An ICC in a color managed system reads the working space you're using and matches the print output to that space, which usually contains non-linear segments, like the compressed shadow values of the Gray Gamma 2.2 space.  Again, however, a key part of making the ICC is to have a feedback loop where the ICC "learns" the characteristics of the printer and inkset, and is then able to adjust for that to make the "linearized" output match the monitor view, in this case no matter what gray working space you're using.

When I'm using Gray Gamma 2.2 working space and printing with QTR, the output from QTR will not match the view on the monitor.  So, I have to adjust for that.  The QTR download package has some ICCs that can be used to tag the files so that the view on the monitor will match the output.  (I use another method -- a Photoshop curve -- to accomplish that.)  Even though QTR's literally linear output does not match my working space, having a "linearized" QTR profile is critical to allowing me (with my adjustment curve method) to consistently have prints that match my monitor rather well.


Hope this helps.

Paul
www.PaulRoark.com