Profiles And Work Spaces, Tagging And Embedding

A profile can be thought of as a set of values which determines how the image pixels are interpreted, either on the screen in the case of the front end profile, or in the print in the case of the back end profile.  They affect the brightness and contrast of the image and are similar in effect to the adjustment curves we place on image layers in the sense that a curve takes the value of a pixel (input) and displays it at another value (output).  So for our simple purposes here we can think of a profile as simply a curve applied in a certain manner.

An image needs a profile in order to display correctly on a monitor, so there is always a default front end profile which is assigned to an image when it is converted to BW from color or scanned in grayscale.  This default profile is set in the Edit/Color Settings window by selecting it from the Working Spaces/Gray picklist.  Dot Gain 20% is the one I like to use for the default for several reasons beyond the scope of this article, but briefly, I have found that it produces the best quality starting print in terms of density, contrast and shadow separation. 

This default profile is called the program's  "working space".  When a BW negative is scanned and imported into Photoshop, or a color digital image is converted to BW, the working space profile is automatically assigned to the image, and it is then referred to as the "document space".  Assigning this profile is called "tagging" the image, and at that point the document space and the working space both have the same profile.  It is possible to change the document space by assigning a different profile, without changing the program's working space.  It is possible to have multiple images open, each with its own document space.

When the image is saved the profile is saved in the file with the image (if we choose to allow it).  This is called "embedding" the profile.  It is important to note that embedding a profile does not change the pixel values in the image.  Apparently there is a reserved space in the file which contains the plot points for the curve.  The curve is used merely to interpret or "map" the image as it is displayed.  There is another menu option to "Convert" the profile, in which case the image pixels are actually changed.  This is used for other purposes beyond the scope of this article.
 

Assigning Front End Profiles

Different front end profiles may be assigned to an image at any time with the "Assign Profile" window

    (PC  Menu: Image/Mode/Assign profile...) 

This window contains three choices:

• Don't Color Manage This Document
  
  
• Working Gray: ...
  
  
• Profile: ...
  

The 2nd choice, Working Gray, is always followed by the name of the current Working Space/Gray profile as specified in the Color Settings window.  The 3rd Choice, Profile, contains a picklist which allows us to choose any of the available gray scale profiles.

Choosing a different profile immediately changes the appearance of the image on screen, so you can see the result of the choice before confirming it and closing the window.

The 1st choice, Don't Color Manage, is different.  This choice removes the assigned profile from the document space and the image is now said to be "untagged".  The significance of being untagged is that if the image is Saved to a file there will be no profile imbedded.  As for the appearance on screen, when the image is untagged the active front end profile reverts to the current working space.  If the profile was already the same as the working space the image will not change when untagged.  If it was something else, then the image will change.  The image appearance on screen always remains unchanged when going back and forth between the 1st and 2nd choices.

It is important to note that any changes made in this window are temporary.  They only affect the image in memory.  The actual file is not changed until the next Save.  When the "Save" command is called, whatever profile is assigned at that time (if one is assigned, or "tagged") will be automatically embedded in the file.  If the file is being saved for the first time, the "Save As" window opens, which contains a check box labeled "ICC profile:", followed by the name of the currently assigned profile.  This gives you the option of embedding the profile or of saving it as an untagged image.

I recommend embedding it because you may end up using different profiles for different images, and you may not remember which profile was used on a given image.  Since it can be changed at any time, and since tagging an image does not change its pixel values, there is no risk in embedding the profile and you are not locked into it forever.

 

The Back End Printer Profile

Another profile is assigned in the Print With Preview window, as described in Part 3.  This printer profile, referred to as the "back end", reinterprets the image pixels on the way to the printer.  At this point things begin to get a bit confusing and the logic to explain it all is somewhat circular.  For the moment, it is important to know just two things:

1) We only use this profile to change the printed output to match a certain condition, such as a different paper.  For example, if we work up an image using EEM for proofing, we may find that it looks a little too light when we print to a brighter paper such as PR, and the degree of difference may vary from one image to another.  By applying a different back end profile we can adjust the PR print to match the proof without altering the image itself.  We also have the ability to print on many different papers by applying different back end profiles.

2) The way to not change the printed output is to be sure that the back end profile is the same as the front end.  This is why I recommend selecting "Same As Source" from the printer profile pick list (see Note below for Photoshop CS2 version).   This selection ensures that they are always the same, no matter which profile is assigned to the front end.  Otherwise we would have to remember to change both each time.

Now hold on to these ideas and bear with me as I lead us on a little  magical mystery tour.  Then we'll come back to this point and discuss different ways to use back end profiles.
 

The Dot Gain Profiles

The concept of dot gain comes from the world of printing presses, where ink dots spread to varying degrees depending on the type of paper, which of course affects the appearance of the image.  All of the Photoshop literature I have read describes the Dot Gain profiles in terms of preparing an image for the press, where the print shop will tell you what the dot gain is for the paper they will be using and you apply the appropriate profile to adjust for it. 

This dot gain concept does not apply to ink jet printing on coated papers. We are not given dot gain ratings for our papers.  In addition, we are not sending our files to be printed elsewhere.  So theoretically the dot gain profiles are irrelevant for our purposes.  However, they can be very useful in a different way.  In general terms, the profiles simply lighten or darken an image by various amounts, and also change the contrast to a small degree.  In order to have control over all this we must understand how the front end and back end profiles interact.
 

Back End Vs Front End

We are provided with seven gray scale profiles to choose from: two Gray Gamma (1.8 and 2.2) and five Dot Gain (10, 15, 20, 25 and 30%).  Any of these can be applied to either the front end or the back end, and they perform a sort of tug-of-war on the image.  A good way to illustrate this is to do some experiments with the five dot gain curves.

Note: If you want to do this yourself, I recommend downloading Paul Roark's "Enhanced 21 Step Wedge" file from the BW Forum's file library.  It provides both a step wedge and a smooth gray scale, as well as numbered patches for a calibrated scale.  Any image can be used to illustrate the point of this article, but this step wedge is useful for many things and is well worth the effort to download it.

As we have already seen, the screen image changes when we assign different front end profiles.  As you select each of the dot gain profiles, notice that the image lightens as the dot gain number gets smaller, and vice versa (the idea here is that a paper with a small 10% dot gain will make a very light reproduction.  The light profile makes us darken the image to compensate).  Something quite different happens when we apply these curves at the back end, however.  Let's follow these steps:

1) Begin by assigning Dot Gain 20% (DG-20) to the front end.  At the back end assign the same DG-20 (or leave it set to "Same As Source"), and make a print (small quarter page prints are fine for this).  This will be our reference print.

Note: In Photoshop Version CS2 these settings are done differently. To set a specific printer profile you must first choose "Let Photoshop Determine Colors" in the "Color Handling" picklist, and then choose the desired profile in the "Printer Profile" list.  To get the equivalent of "Same As Source", where the back end is automatically the same as the front end, select "No Color Management" in the "Color Handling" pick list.  The Printer Profile list will be disabled.


2) Now assign DG-15 to the back end (leave the front end at DG-20) and make the print .  It got darker!  Didn't we establish that smaller numbers got lighter?

3) Now assign DG-25 to the back end and make the print.  It's lighter than the reference print.  If you were to continue and make prints using DG-10 and DG-30 you would see the same changes in greater degree.  What's going on here?

This is an important point -  The Dot Gain profiles used on the back end have the opposite effect as when used on the front end.  Why?  I'm not sure but it doesn't matter.  As long as we can figure out how it works we can use it to our advantage.  Let's explore some more.

4) Assign DG-10 to both the front and back ends and make a print.  It's the same as the reference print!  Very strange.

5) Assign DG-30 to both the front and back ends and make a print.  Again, the same.

What's happening is that a curve at the back end applies exactly the opposite effect as it does at the front end.  When the same curve is applied at both ends, the back end curve is canceling out the front end curve altogether and the image defaults to a standard output of some sort (more about that later).

This brings us to another important point: As long as both ends have the same profile the print will look the same, no matter which profile is used.

If we continue to experiment by assigning various different profiles to the front and back ends, we would discover yet another important point: Every possible combination of different front and back end profiles will produce a different print.

Note: I recommend trying this for several combinations using the step wedge mentioned above, just to get a feel for how the profiles interact.  The gray scales in the images will show clearly how they are affected.

This presents a real problem because any specific back end profile, no matter what it is, will react differently with various front ends.  So if you change front ends, not only does the screen image change, but the print also changes, often in ways which make no sense.  It can be very confusing and you can feel like a puppy chasing its tail around in circles.  Therefore, using front end profiles as a tool for manipulating the print is unintuitive and inefficient.  It would seem that the only way to have consistency in the work is to set both ends at some particular combination and never change it.  But that is rigid and confining.  Is there not some way we can use front end profiles to our benefit?
 

The Search For WYSIWYG

If we can achieve WYSIWYG, where What You See (on screen) Is What You Get (in the print), then our work will be more efficient and require less time, ink and paper while working up our images.  Getting true WYSIWYG, however, is easier said than done, but we can use front end profiles to help us get closer. 

Remember that if we keep the same profile assigned to both the front and back ends, the image always prints the same - changing front ends has no effect on the print.  This provides the consistent output we need, and allows us to change the front end profile to make the screen image match the print.  By simply setting the back end to "Same As Source", we can experiment freely with different front ends until we find the one that best suits the image.  For example, a high key image might work well with Gray Gamma 1.8, while a low key image might have better WYSIWYG with Dot Gain 20%

So here we have a completely different approach:  At the front end, the profiles are a tool to help us achieve WYSIWYG as we work up a print on our standard proof paper.  At the back end they help us match the completed image to a particular paper for the final print.
 

Using Back End profiles

When the time comes to make a final print on another paper, the print may appear too light or too dark, or the contrast may be different, depending on the papers involved.  We can use the back end profiles to adjust the image to match the proof.  This came in handy for me once  when printing a snowy winter landscape on PR.  I had worked up the image on EEM using a Dot Gain 20% front end.  The large high value snow area, which was sensitive to slight changes, looked a bit weak on PR because it is a brighter paper.  Choosing Dot Gain 15% for the back end darkened the image to look almost exactly like the EEM version.

This simple approach is often all that is needed, but sometimes a finer degree of adjustment is required.  The default Dot Gain curves of 10, 15, 20, 25 and 30% give us a rather crude scale of 5% jumps.  We gain a little bit more flexibility at one end when we add the Gray Gamma curves to the mix.  On a brightness scale, GG 1.8 is between DG-20 and 25, an equivalent of about 22, and GG 2.2 is between DG-25 and 30, at about 28 (these are eyeball estimates, just to give an approximate idea).

Sometimes these choices are too crude for a brightness adjustment, and they may also alter the contrast unacceptably.  When this happens we can create customized curves which can be saved for re-use.
 

Custom Back End profiles

The ability to create custom back end profiles is another feature in the "Show More Options" section of the Print With Preview window.  If you select "Output" in the options picklist, the bottom section changes and several buttons appear.  Click the "Transfer" button and the "Transfer Functions" dialog opens.  This allows us to adjust the density and contrast of the output with a fine degree of control. 

The easiest way to lighten or darken the print is to click on the 50% edit box and enter any number greater than 50 to darken, less than 50 to lighten (in this grid the lower left corner is white).  This moves the mid point of the curve up or down, and the entire image shifts slightly, but without moving the low and high anchor points. 

 Note:  This is an important point.  This brightness adjustment is not linear.  A linear adjustment, such as the Brightness/Contrast Sliders in the Image/Adjustments menu,  moves every pixel up or down by the specified amount.  If you brighten by a value of 5, for example, any pixels which are 0 become 5, and you lose your blacks.  Any pixels which are 250 or higher become 255, so you blow out the highlights.  The reverse happens when you darken.  For this reason the experts say never to use this slider control on an entire image, but reserve it for selected areas of limited range.

Back to our curve, moving the center point while holding the low and high anchor points preserves our blacks and highlights, but it also means the slopes of the curve will change.  For example, darkening the mid point means the slope on the dark side of the mid point will be lessened (decreasing contrast) and the light side slope will be steepened (increasing contrast).  The more the mid point is moved, the more noticeable these changes will be.  

In practical experience I have found these changes to be insignificant when darkening the image slightly to adjust for PR paper (a recent image required a darkening of 1.57%, equivalent to 4 RGB units).  Nevertheless, more data points can be edited in the curve if a contrast adjustment is needed. These values are accurate to a hundredth of a percent which gives us an extraordinary degree of control.  The curves can be saved and reused.

 

Custom Front End Profiles

In the attempt to find the best WYSIWYG front end for an image we may find that we need a curve somewhere between the available choices.  Suppose, for example, we find DG-20 a bit too dark, but DG-15 is much too light - we need something around 18%.  Since the profile picklist in the Assign Profile window doesn't have this choice, the only way I have found to allow this is to use a custom dot gain curve for the Gray work space in Color Settings.  This will then appear as the "Working Gray" 2nd choice in the Assign profile window.  Here's how to do it:

1) Open the Edit/Color Settings window.

2) Open the Working Spaces/Gray picklist (currently set to Dot Gain 20%) and select "Custom Dot Gain".  This will open a Custom Dot Gain window with a curve grid very similar to the Transfer Function grid used for custom back end profiles.

3) Notice that the grid already contains a curve, with the 50% point set to 70.  This represents the Dot Gain 20% curve that is the current setting.  The 20% means the degree of gain at the mid point.  Since the straight line value of the midpoint is 50, then 70 is equal to a 20% gain. So all we need to do for a custom curve is to add the required % to 50 and enter that number in the 50% edit box.  So for our 18% we need to merely change the 70 to 68.

4) We must name our custom curve, so in the Name edit box add "18%" to the default "Custom Dot Gain" name.  Click "OK" to confirm and close the window.

5) We must save this as a custom color setting file (.csf), so click "Save" to open the Save window and enter a File Name that will have some meaning later, such as "Gray Dot Gain 18".  Click "Save", then enter a descriptive comment in the "Color Settings Comment" window, then click "OK" to confirm.

6) Now open the Settings picklist and this file name should appear in the list, where it is always available.  Click "OK" to confirm and close the Color Settings window.

Now the Assign Profile window will show this as the current Working Gray and we can assign it to the image.  These custom color settings can be created and saved as needed, and over time you will have an entire set of finely graduated curves.  It is a bit tedious to have to change the Color Setting just to assign a particular curve, but if you are going to spend hours working on an image it is well worth it to get better WYSIWYG. 

Note: The Assign Profile window picklist has a grayed out item called "Other", which implies that there may be a way to add custom curves to this list.  As of this writing I have not found a way to do this.  If anyone figures it out please let me know.
 

The Printer Driver

Here we are entering a mystery zone where things are happening which I don't fully understand, but I can show the results of some experiments which will shed a bit of light on the subject.  I stated earlier that when the front and back end profiles are the same they cancel each other out and the output reverts to a default setting of some sort.  I believe that default is the color management provided by the printer driver.  The descriptions that follow are from my Epson 2200 driver.  These settings may vary with other printers or drivers.

I often print using the Black Only technique, and when I check Black Ink Only in the settings window all of the Color Management choices are grayed out except one called "Color Controls", and it includes a picklist with three gamma settings, 1.5, 1.8, and 2.2.  These settings represent three different profiles which determine the overall density of the print.  2.2 produces the darkest print, and 1.5 is the lightest.  There is no way to not choose one of these.  I keep mine set on 1.8 because this produces a print that is closest to the density I want (using the other settings would require changing the image to get the density that 1.8 provides already - why bother?). 

Note: The new Epson K3 printers in ABW mode don't have this driver gamma setting, but do have a "Tone" setting with "Darker, Normal, and Lighter" choices, which seem to perform the same function.

So the bottom line for consistent printing is to choose a driver gamma setting and leave it alone.  Using it as another print control just adds an unnecessary degree of complexity.
 

Summary

Conclusion

Copyright © 2002 - 2005 Clayton Jones
All rights reserved.  This article may not be reproduced without express permission of the author