This is an update to a problem where a digital RGB file was converted 
to the (printer's suggested) US Web Coated (SWOP) V2 CMYK profile for 
printing as a postcard. Over the print run the neutral background 
colour did not stay neutral but varied unacceptably to three 
different shades from geenish through neutral to magenta-ish. In an 
effort to understand why, I put the question to the list before 
contacting the print company. Stephen Marsh has worked with me in the 
most helpful and instructive manner, to find out what the problem was 
and see what could (have) been done to avoid this end result.  Our 
correspondence (condensed) follows.

Several people contacted me offlist and asked for this, and the 
acting Listmum suggested that I post for the benefit of all 
interested.

His actual action is not included, though his explanation will be 
enough, I'm sure. 

The image - RGB, reduced in size and converted to jpg for the web - 
is to be found here:

http://www.iistudio.com/cmyk_problem.html
(copyright Steven Kennard 2004) 

It appears that the fault is divided between the print company and 
myself. In my case it was ignorance as to the fact that this was an 
unusual image that needed special treatment. This particular image 
needed two different types of separations to allow the pressman to 
hold a neutral.  

Stephen Marsh said:

SM: If a neutral tone is formed mostly of CMY with little K ink - it 
is harder for the pressman to hold a neutral. On the flipside, using 
more K in a coloured job can also have issues, but in neutral grey 
areas it is often a good thing.  

This job could have used higher GCR than is offered in the SWOP v2 
profile (TR001 colourimetric data). This uses  medium type GCR in 
this Adobe profile.  

I see two issues with this image:

a) The background which should be neutral.

The CMYK file you supplied has very good and traditional neutral 
CMYK values, as in the CMY are good for greybalance for TR001 
conditions...if the printer can hit these conditions with the 
stock/ink etc. It sounds like they did in 1/3 of the cases getting 
corrct neutrals and correct brightness.  

b) The main image itself (not the background)

You would probably need to separate the "hero" differently than the 
background. In essence this image can greatly be enhanced by using 
two different separation methods and combining them, taking the best 
results of each conversion and masking them together. A heavy GCR 
separation for the gradated netural background and a lighter GCR for 
the main image content with special attention to the K plate.  


With neutral grey balance on average being formed from + 50c40my 
with 10K - the bias is on getting the CMY correctly in balance for 
the background. More K in the background would have been better, but 
NOT more K in the image! Thus combining two separations into one.  

EK: Were my expectations too great?

SM: Yes and no.

For the GCR ratio that you supplied and the image content - yes.

In your general expectation - no.

It needs to be understood that what the aimpoint is...have you 
communicated your desires correctly and are they understood? Is the 
price of the job going to warrant first class work? Some printers 
gang up multiple jobs to save costs (bulk business card or post card 
printers etc) - your job is one of many on the entire press 
sheet...due to the mechanics of offset print large amounts of colour 
can affect the draw/flow of ink to other areas of the press sheet. 
The colour design of elements on one page or the inclusion of other 
pages may alter the content in other areas. 

Is it a neutral background in all cases at the expense of the main 
image? Is it the main image at the expense of the background? Should 
both the main image and the background be reproduced ideally?   

Prepress is my trade, I have been doing it for the last seventeen 
years. I find it sad that now that anyone can do their own prepress, 
my craft has been devalued a lot. For many images and jobs, an 
acceptable separation can be performed with little forethought of 
the original and the destination and a simple push button approach 
can be performed. But some other images are best handled with 
experience and human analysis - which is missing from a profile 
transform which does not know anything about the image.  

EK: I shall go to Dan Margulis' book (the only one I have that deals 
mostly in CMYK) and see if I can make more sense of it.  

SM: He has a section on this about GCR and neutral tones under 50% 
brightness...which need heavier GCR, just like I am talking about 
and demonstrating with this action and emails.  


EK: Does this mean that, in your opinion, I should say nothing to the 
print company and realise that this was entirely my ignorance on how 
to correctly treat this image for press? Perhaps this is just a 
learning experience for me?   

SM: Hmmm...

They did recommend a profile which did not deliver enough GCR for 
the image content to give the press operator a reasonable chance of 
holding consistent neutrals and tonal brightness over the print run. 
If they saw the image first and they knew their game, they would 
have noted that the image is not typical and perhaps should not be 
handled typlically - or that different expectations may be set 
between all parties etc.  

So, if you can get a 2/3 or 1/3 free reprint of the existing 
data/separations to match the prints that you are happy with...then 
that is good. Also chalk it up to experience as well. Both parties 
have equal blame/responsibility in my opinion.   

If they are willing to learn from this as well as you, they might 
wish to see what a new separation provided by you prints like using 
methods suggested my myself...this may educate them and yourself and 
provide a good test piece or education piece etc. You might be able 
to negotiate something here.  

Presswork really is a crapshoot, a gamble. The idea of the 
knowledgeable prepress operator is to second guess the press issues 
and provide a separation which will aid the press operator. The 
problem is that it can be hard to get a good answer from press guys 
as to what is good or bad until ink hits the paper - many are good at 
evaluating files or separations but many are not and prepress is 
often more experienced at these decisions. 

The steps he used to get the image optimised for the press follow:

* A base conversion is performed to generate the correct separation 
for the hero or foreground object. In this case I used the Chromix 
UCR TR001 profile instead of the Adobe TR001 SWOP v2 profile as I 
prefer the conversion for this image as it has a contrast kick in the 
perceptual intent whereas the Adobe one does not (this is not too 
critical). Using either TR001 profile would be OK here.  

* A dupe of the RGB is also made and converted using Photoshop's legacy
Custom CMYK engine using special custom values to generate a heavy GCR
separation which is close to TR001 aimpoints. By default custom CMYK does
not match the same grey balance as TR001 even though it has SWOP in it's
name. As one can't edit the Adobe v2 or Chromix profiles in Photoshop to
create heavier GCR one must use the old custom CMYK to generate the required
amount of black in the four colour grey neutral background.

* Both CMYK files have the same SWOP v2 profile assigned to them, as they
are both separated for the same print condition (they just happen to use
opposite black generation methods!!!).

* One CMYK file is layered over the other. One then masks the image so that
the file with heavy GCR has the foreground hero image hidden/masked. This is
the upper layer above the lighter UCR separation that is good for the hero
image.

* A mask is automatically generated based off grey density using the layer
option blend if sliders and also using a layer mask generated from the K plate.

(Stephen then corrected himself with a second action and this 
explanation:)  

I was not happy with my layer option blend if sliders used to mask 
the image - the manual layer mask is better but must be refined by 
hand. This new action resets the upper layers blend if settings to 
default so that they do not contribute to the masking effort...now 
only the layer mask is used  
and the layer mask obviously needs to be tracing the foreground 
object well.

Many thanks once again, Stephen!

Ellie
-- 
Ellie Kennard, Innovative Imaging Studio : http://www.iiStudio.com
Contributor to "Photoshop Restoration and Retouching, 2nd edition"
by Katrin Eismann


===============================================================
GO TO http://www.prodig.org for ~ GUIDELINES ~ un/SUBSCRIBING ~ ITEMS for SALE

Reply via email to