[scribus-dev] GSoC project status: PDF/X export

[Pierre Marchand]

Vous avez écrit :
> 2009/6/15 Pierre Marchand <pierremarc at oep-h.com>
>
> > Vous avez écrit :
> > > As for PDF/X-1a, since this format enforces blind exchange where all
> > > colors need to be in DeviceCMYK, I have decided to let users export to
> > > PDF/X-1a only if the Color Management is turned *off* in the
> > > document's setting as opposed to it to be turned on in exporting PDF/
> > > X3.
> >
> > Bad. Having unmanaged (generic in fact) color transformation is not
> > exactly the first step of a controlled printing process.
>
> I am not too sure wether this is good or bad since PDF/X-1a requires
> explicitely no RGB data but CMYK data. One could assume that the CMYK files
> have been produced in a controlled environment and contain the necessary
> data.

Here, the "controlled environment" is Scribus itself. If you ask it to produce 
a CMYK file, it will convert colors of objects from their declared color space 
(either on a per item basis or taking it from document’s settings) to the 
target color space specified for the document. Or, if you disable CM, it will 
do the very same conversion but based on an _implicit_ target profile.
Jean can correct me if i’m wrong or imprecise.

>
> What strikes me more here is the fact that we would claim that Scribus
> supports PDF/X-1a but at the same time the application is, afaik, unable to
> flaten the transparency eventually used in that file. In fact, this will
> make the format unusable for just any Scribus document containing
> transparency.

Whereas it can’t flatten transparencies, it can, and does, check that no 
transparency is used in a document before exporting to PDF1.3.

>
> I know we have discussed briefly this on IRC and it has been mentionned
> that flatening transparency is a complicated task. But shouldn’t one come
> with the other?

Not necessiraly. Writing a flatener will/would be a very long run job. And, in 
my opinion, it can’t be handled by only one person in one shot. I expect it to 
be written little by little, on an incremental basis, as ideas come.
In the meantime (and this kind of can late :-)  I think the actual strategy of 
checking tranparency in objects before exporting is enough. Note that when 
we’ll un-mark PDF embedding as experimental, we’ll have to do a lot more 
checks on embedded content streams. Examining both the PDF version, graphic 
state, etc.


-- 
Pierre Marchand
http://www.oep-h.com