Best common file format to use to create PDFs?

> "Zak" <duff RemoveThis @nomail.invalid> schreef in bericht
>>
>>I use XP.
>>
>> I scan a document to a graphics file (eg jpg, gif, bmp, etc)
>> Then I make a PDF from the graphics file.
>>
>> I see that if I save a large bitmap (BMP) and then create a PDF
>> then the PDF is no bigger than if I had compressed the image to a
>> GIF and used that to create a PDF file.
>>
>> In order to preserve quality, avoid artefacts and interference
>> patterns, would it be better to save to a large and detailed
>> intermediate file like a BMP (or even a jpeg) or to save to a
>> small lossless file like a GIF or TIF?


On 26 May 2006, Nils<bla RemoveThis @bla.com> wrote:
>
> Afaik, GIF is not used as a graphical format inside PDF. It is
> probably compressed as a "TIFF" with Cittfax or LZW compression.
> "TIFF" is between brackets, because the embedded stream is also not
> a complete TIFF file, just contains the compressed graphics and
> some extra information like scansize, colordepth, color channels,
> etc.
>
> What happens under the hood in your PDF creation really depends on
> the PDF engine you're using. Many engines actually resize your
> graphics to match the PDF DPI resolution. If you're an experienced
> programmer you could try to generate the PDF yourself, with the
> images in full resolution. The PDF specification is open and can be
> found on the Adobe website.
>
> Nils


Hi Nils and others. I understand now that when I create a PDF from a
image file that the format of the image file is not used inside the
PDF. Instead some other format is used in the PDF (which Nils kindly
suggests may be a specialized form of TIFF).

It is this conversion from my image file format to the internal PDF
format which I want to be done smoothly. I am on XP and I am
wondering if it is better to start with a GIF or a JPG or BMP or
whatever to feed into my PDF creation utility.

I should say that I am starting with a hard copy of a document
created on a word processor. I want to avoid artefacts, unecessarily
jagged lines, moire effects and all that stuff which might come from
transforming from an "awkward2 image format to a PDF.

My PDFs will be for public distribution. I have preferred to scan to
a GIF file rather than a TIFF because I have assumed that when I
circulate the basic image file among certain people that the best
balance between image size and the best chance of them being able to
see the file is a GIF.

To me TIFF feels a bit specialized. For example, I never see a web
page with TIFF images but I see lots of pages with GIFs.

Also there seem to be various compression options for a TIFF (group 3
or 4, LZW, JPEG deflate, none) which might makes it even harder for
me to know what to choose as a common format! The Wikipedia says
documents are often scanned to TIFF group 4 but is that something
which has the best chance of being seen on various PCs in various
organisations that I might need to send it to?

["Followup-To:" header set to comp.periphs.scanners.]
On Tue, 30 May 2006 16:30:29 +0100, Zak staggered into the Black Sun and
said:
>> "Zak" <duff.DeleteThis@nomail.invalid> schreef in bericht
> On 26 May 2006, Nils<bla.DeleteThis@bla.com> wrote:
>>> In order to preserve quality, avoid artefacts and interference
>>> patterns, would it be better to save to a large and detailed
>>> intermediate file like a BMP (or even a jpeg) or to save to a small
>>> lossless file like a GIF or TIF?
>> Afaik, GIF is not used as a graphical format inside PDF. It is
>> probably compressed as a "TIFF" with Cittfax or LZW compression.
>> "TIFF" is between brackets, because the embedded stream is also not a
>> complete TIFF file.
>>
>> What happens under the hood in your PDF creation really depends on
>> the PDF engine you're using. Many engines actually resize your
>> graphics to match the PDF DPI resolution. If you're an experienced
>> programmer you could try to generate the PDF yourself

Zak is obviously not a programmer, let alone an experienced one. Using
PDFlib from C isn't that difficult if you have some experience in C,
though.

>> The PDF specification is open and can be found on the Adobe website.

And creating a PDF using only the spec would take a bunch of experienced
programmers a while. The PDF spec is really, really complex. Its
complexity is one reason why PDFlib and ps2pdf and OpenOffice's "print
to PDF" functionality exist.

> I understand now that when I create a PDF from a image file that the
> format of the image file is not used inside the PDF. Instead some
> other format is used in the PDF (which Nils kindly suggests may be a
> specialized form of TIFF).

Using tiff2ps -> ps2pdf says that a grayscale TIFF ends up converted to
a stream object that can be decoded by the FlateDecode filter.
YPDFEngineMV, obviously.

> It is this conversion from my image file format to the internal PDF
> format which I want to be done smoothly. I am wondering if it is
> better to start with a GIF or a JPG or BMP or whatever to feed into my
> PDF creation utility.

Depends on what you want. Get a good scan, and convert it to
black-and-white if you can do that without losing important info;
that'll make the PDF smaller. JPEG may introduce artifacts, so you
probably don't want to use that. TIFF G4 and TIFF LZW are lossless, so
you may want to use those.

> I should say that I am starting with a hard copy of a document created
> on a word processor.

Yuck. The original WordPerfect or whatever file would've been a much
better place to start from. PDFs with just text in them tend to be
smaller, display faster, and can look good at any zoom level. PDFs made
from images take a longer time to display, are larger, and look terrible
at high zoom levels.

> I have preferred to scan to a GIF file rather than a TIFF because I
> have assumed that when I circulate the basic image file among certain
> people that the best balance between image size and the best chance of
> them being able to see the file is a GIF.

? You're creating a PDF, not distributing a series of image files.

> To me TIFF feels a bit specialized. For example, I never see a web
> page with TIFF images but I see lots of pages with GIFs.

This is because of Hysterical Raisins in the history of web browsers,
and because of Unisys being asses. JPEG compresses better than TIFF-LZW
for lossy color images, and smaller images are preferred, especially when
you're on dialup. TIFF-LZW gives the best lossless compression for
color images, but TIFF-LZW is usually used where losslessness is more
important than file size (like in prepress.) Also, Unisys said they'd
sue anyone who made a TIFF-LZW compressor unless they paid Unisys a
license fee.[0] These things combined made it so that the earliest GUI
browsers didn't support viewing TIFFs, just JPEGs and GIFs. And this
has persisted to the present day... even though TIFF-G4 compresses
better than *anything* else, and does so losslessly, iff your image is
black-and-white.

> Also there seem to be various compression options for a TIFF (group 3
> or 4, LZW, JPEG deflate, none) which might makes it even harder for me
> to know what to choose as a common format! The Wikipedia says
> documents are often scanned to TIFF group 4 but is that something
> which has the best chance of being seen on various PCs in various
> organisations that I might need to send it to?

....what? If somebody can't figure out how to view a Group4 TIFF,
they're probably computer-illiterate. Anyway, aren't you making a PDF
here? It doesn't matter what the original image format was if it's been
PDFed. Acrobrat Reader can decode the image data within a PDF, as long
as the PDF library/PDF writer/whatever that created that PDF wasn't
smoking crack. Anyway, HTH,

[0] Fortunately, their patent (on a *mathematical method*!) expired a
couple of years ago, so all the Free stuff can write LZW now, which is a
win for everybody.

--
Matt G|There is no Darkness in eternity/But only Light too dim for us to see
Screw up your courage! You've screwed up everything else.