[gs-devel] Colour Separation output

[Ken Sharp]

At 13:12 10/01/2008 +0000, you wrote:


>Does this mean that for true CMYK + Spot files the outputs would be 
>correct and for RGB, I would need to do these extra calculations to get a 
>best guess for the CMYK?

Yes and no. I believe that if your job was CMYK+spots then it would 
generate the output you expect. I don't think that you need to adjust UCR 
and BG though, I think the device in question is 'mapping' RGB to CMY for 
you, rather than converting to CMYK. As a result the UCR and BG isn't being 
applied.


>If so then it is the RGB files that is confusing.
>
>I am not an expert but am aware of the UCR etc which would affect the 
>values, but not sure of the calculations.
>
>in your calculations, is:
>
>UCR(K) = min(C,M,Y)

That's a very simplistic calculation, just intended for an example, it just 
selects the smallest of CMY and removes that from each of the CMY channels 
and calls it black. In practice there may well be better calculations, and 
also that's not actually how the UCR function gets called. But I think this 
is a red herring right now.


>and when you mention
>"... depending on what you have UCR and BG set to"
>does this mean the 'minimum' value can be tweaked for some kind of 
>calibration?

The UCR and BG can be adjusted for a primitive form of colour correction, 
its the only way to affect the RGB->CMYK conversion in the PostScript 
language (rips often supply other proprietary techniques as well, but these 
aren't part of the specification).


>Also, how would I determine if the file had RGB elements in it, in order 
>to use this best guess scenario?

Well... If the file doesn't have RGB then the UCR and BG won't get used 
(when printing to a CMYK device), so they don't do any actual harm. If you 
want to know whether a PDF file uses RGB then you can use the Acrobat 
preflight tool, just as I did.

For PostScript (and automated detection of RGB in PDF) this is a great deal 
more difficult and involves using an interpreter to check the content of 
the file before rendering. Probably not worth it in general.


But I think the 'issue' is the way the tiffsep device works with RGB files. 
It seems that it maps RGB straight to CMY rather than doing a conversion to 
CMYK.

This looks deliberate, and I'm guessing its done so that a single device 
can produce RGB, CMY and CMYK separations. I haven't tested it but I 
imagine a DeviceGray would map straight to black, so the device can support 
any of the basic colour models and probably any of those with additional 
spots as well, but I haven't tested that either.

I'm hopeful that Ray or one of the other engineers can give you more 
guidance than I can here, its not my area and there are an awful lot of 
switches and controls I don't know anything about. Maybe the device can be 
reconfigured to do the conversion to CMYK for you.



                                             Ken

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
TASK: Shoot yourself in the foot.
Prolog (interpreted) Your program tries to shoot you in the foot, 
but you die of old age before the bullet leaves the gun.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -