freetype-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Devel] Re: FreeType -> PS/PDF


From: Leonard Rosenthol
Subject: [Devel] Re: FreeType -> PS/PDF
Date: Mon, 24 Dec 2001 18:02:18 -0500

At 05:37 PM 12/24/2001 +0000, Vadim Plessky wrote:
Will the license be compatible with FreeType and/or BSD?

My rewrite would use the FreeType license, since I'd add it to the FT CVS repository (assuming David doesn't mind ;).


What kind of license your former employer is using?

        None, it's proprietary.  Part of the reason to rewrite it.



Embedded font is PS Type3 font, so all TrueType hints are lost, and there are
no PS hints generated.

Yuch! Not only don't you have hints, BUT you also get (as you noted) really crappy screen display.

The Type3 conversion hack is quite common in the Unix PS generation community - probably going back to the need to do that with TeX's MetaFonts. It is also a "cute hack" to provide subset embedding w/o actually implementing real subsetting.


[NOTE: I remember that David Turner wrote some time ago that Acrobat 5.0 uses
new different algorithm for auto-hinting of PS fonts, and ignores hints for
embedded fonts; I haven't tested Acrobat 5.0 and can't confirm this, but
overall it seems quite realistic and reasonable ]


News to me if it's true...Acrobat (and Ghostscript and Xpdf) all uses the embedded data (if present, of course) to display the font. So if the hinting data is present, it will be used. Of course, it is certainly possible to skip that data when embedding it, especially if doing subsetting.


It worth to mention here that, according to that thread and some posters,
some PDF-generating software positions *each glyph individually*, therefor
you can't just "copy and paste" text - as it's required by WCAG (Web Content
Accessibility Guidelines)

        The two are NOT related.

It is certainly true that some PDF generators position glyphs individually - especially ones doing "high end" typographical functions like kerning or tracking. HOWEVER, that doesn't mean that you can't copy & paste.

When Acrobat and Xpdf display a PDF, then do some pretty complex stuff to reorganize the physical (in file) ordering of the glyphs into the logical (on screen) ordering so that text selection & extraction happens correctly - regardless of the source data. Ghostscript's text extraction does similiar things, though not as complex since it doesn't need to worry about "live selection".



So, embedding some text in PDF (and/or SVG) doesn't offer you warrantry that
you can retrieve text later (for reading via some Screen Reader, for example)

If does for a properly written tool. In fact, Acrobat 5 for Windows directly integrates with Screen Reading software to present it with logically ordered data that it can use. It will also use, if present, structure information so that a table, columnar text or other complex layout is read in the logical order for the user.


and, as I mentioned above, for Accessibility as well.
I am not big expert in Accessibility so far (but working on improoving my
skills), but if we finally come to some open-source SVG-to-PDF or
2D-graphics-with-text-to-PDF generator, I think we should take care about
blind people or people with visual disabilities in advance.

        Agreed!

The way to do this for PDF is to use the new features of PDF 1.3 and 1.4 to add "structure" and "tagging" to the file. This is what then allows screen readers to read the file in the best possible way.,


It will not cost a lot of extra code but can make thousands, if not millions,
of people happy!


Doing structure and tagging isn't rocket science, but it will require some interesting code. Which is why you don't see anyone (other than Adobe) generating this stuff - yet...


Still it seems Adobe added several extra libraries for their SVG 3.0 viewer
(I refer here to Linux version of it):
address@hidden adobesvg-3.0]$ ls lib*
libACE.so  libAGM.so  libBIB.so  libCoolType.so  libNPSVG3.so  libSVGCore.so

ACE is the "Adobe Color Engine". I'm not familiar with BIB, but I'll ask around...


Leonard




reply via email to

[Prev in Thread] Current Thread [Next in Thread]