Webmaster forum - Interesting PDF Conversion Problem

Author

Interesting PDF Conversion Problem

TS Moderator

2006-09-24, 6:58 pm

OK. I was producing math tutorial documents for my site using
Microsoft Word and embedded Visio drawings. They were coming out just
the way I wanted them, so I decided to convert them to PDF.

I tried a couple of different PDF converters and they gave me
reasonable results EXCEPT -- I noticed that sometimes my embedded visio
objects get converted to bit maps and sometimes they get converted to
some sort of scaleable graphics. This includes lines, geometric
shapes, AND text.

It seems to be a little bit arbitrary. I can copy an embedded Visio
object and add one more line to it or just apply bold to some text and
suddenly things are being converted to bitmaps.

I can think of a few reasons for this, but I would rather not guess,
and I would rather everthing was converted to scaleable graphics, as
they are beautiful and infinitely expandable as the document is zoomed
larger.

Anyone have a clue about this?

Thanks.

TS Moderator

2006-09-24, 6:58 pm

One more thing -- I tried more than one vendor's PDF Conversion
Program, and got the same result.

Scott Bryce

2006-09-24, 6:58 pm

TS Moderator wrote:

> I tried a couple of different PDF converters and they gave me
> reasonable results EXCEPT -- I noticed that sometimes my embedded visio
> objects get converted to bit maps and sometimes they get converted to
> some sort of scaleable graphics. This includes lines, geometric
> shapes, AND text.

PDF is a complex file structure. You might get better answers if you ask
at comp.text.pdf.

I don't know enough about Visio drawings or your PDF converter to have
any idea why you sometimes get line drawings and other times get
bitmaped images. Someone at comp.text.pdf might.

Ed Jay

2006-09-24, 6:58 pm

Scott Bryce scribed:

>TS Moderator wrote:
>
>
>PDF is a complex file structure. You might get better answers if you ask
>at comp.text.pdf.
>
>I don't know enough about Visio drawings or your PDF converter to have
>any idea why you sometimes get line drawings and other times get
>bitmaped images. Someone at comp.text.pdf might.

You're asking your pdf generator to interpret a Visio object, something it
doesn't know how to do. If you're going to make a pdf file, you will want
to paste a bitmap of your Visio graphic, not an imbedded Visio object.
--
Ed Jay (remove 'M' to respond by email)

Andy Dingley

2006-09-24, 6:58 pm

TS Moderator wrote:

> OK. I was producing math tutorial documents for my site using
> Microsoft Word and embedded Visio drawings. They were coming out just
> the way I wanted them, so I decided to convert them to PDF.

You can't put Visio on the web. Or at least: if you did you'd make it
inaccessible for those who didn't have Visio to view it with.

This issue is the same whether you use PDF as an intermediary or not.
Visio is great to draw things with, lousy for publishing. To make its
output visible you need to export it as either some vector format, or
render it to bitmap(s). Rendering it to too small a bitmap will give
you problems with jaggies.

Personally I'd avoid PDF and I'd use HTML and bitmaps. It's a HTML
website and you need to use bitmaps - keep the rest simple, avoid PDF
that you don't absolutely need. Try using bitmaps big enough to print
adequately and scaling them down client-side in the browser (simple CSS
on the <img> tag). This can be a bit wasteful of bandwidth for
screen-only use, but it's a respectable trick for a print-centric site.
You can even use totally different images for screen and print,
selecting them with @media print {...} in your CSS

If you want to use PDF (maybe it's a "download and save for later use"
sort of situation, then check the quality of whatever generates your
PDFs. _This_ tool will be rendering the Visio content to either vector
or bitmap, and with different qualities. PDF is a complex format with
lots of options for how you generate the "same" content into the "same"
PDF. You can see a lot of difference in quality (and file size)
depending on how you do this.

If you really care about the details, then look at using XSL:FO and
maybe Apache FOP as your route to PDF.

TS Moderator

2006-09-24, 6:58 pm

Andy Dingley wrote:
> TS Moderator wrote:
>
>
> You can't put Visio on the web. Or at least: if you did you'd make it
> inaccessible for those who didn't have Visio to view it with.
>
> This issue is the same whether you use PDF as an intermediary or not.
> Visio is great to draw things with, lousy for publishing. To make its
> output visible you need to export it as either some vector format, or
> render it to bitmap(s). Rendering it to too small a bitmap will give
> you problems with jaggies.
>
> Personally I'd avoid PDF and I'd use HTML and bitmaps. It's a HTML
> website and you need to use bitmaps - keep the rest simple, avoid PDF
> that you don't absolutely need. Try using bitmaps big enough to print
> adequately and scaling them down client-side in the browser (simple CSS
> on the <img> tag). This can be a bit wasteful of bandwidth for
> screen-only use, but it's a respectable trick for a print-centric site.
> You can even use totally different images for screen and print,
> selecting them with @media print {...} in your CSS
>
> If you want to use PDF (maybe it's a "download and save for later use"
> sort of situation, then check the quality of whatever generates your
> PDFs. _This_ tool will be rendering the Visio content to either vector
> or bitmap, and with different qualities. PDF is a complex format with
> lots of options for how you generate the "same" content into the "same"
> PDF. You can see a lot of difference in quality (and file size)
> depending on how you do this.
>
> If you really care about the details, then look at using XSL:FO and
> maybe Apache FOP as your route to PDF.

I guess I disagree. Look at
http://www.weekendsuccess.com/Docs/...FlashCards1.pdf and
http://www.weekendsuccess.com/Docs/GeometryReview1.pdf. I cannot
imagine trying to put those two documents on the web as html documents.
PDF seems great for this sort of thing. However, as I explained, i am
getting INCONSISTENT conversion. Sometimes the PDF documents have
embedded Vectors and sometimes they have embedded bitmaps. I would
like to get some consistency.

BBM

2006-09-24, 6:58 pm

TS Moderator wrote:
> Andy Dingley wrote:
>
>
> I guess I disagree. Look at
> http://www.weekendsuccess.com/Docs/...FlashCards1.pdf and
> http://www.weekendsuccess.com/Docs/GeometryReview1.pdf. I cannot
> imagine trying to put those two documents on the web as html documents.
> PDF seems great for this sort of thing. However, as I explained, i am
> getting INCONSISTENT conversion. Sometimes the PDF documents have
> embedded Vectors and sometimes they have embedded bitmaps. I would
> like to get some consistency.

I don't know about the consistency problem, but looking at those PDFs,
I don't see why they /can't/ be put into HTML format (of course, having
headers and footers in HTML would be less reliable). You can even use
the @media print to make the flash cards the right size on the printed
page while making them look proper on the web.

I'm not saying you /should/ change them to HTML (for your requirements,
I agree that PDF is a suitable medium. It just isn't the only one),
merely that you /can/ change them.

Scott Bryce

2006-09-24, 6:58 pm

TS Moderator wrote:

> I guess I disagree. Look at
> http://www.weekendsuccess.com/Docs/...FlashCards1.pdf and
> http://www.weekendsuccess.com/Docs/GeometryReview1.pdf. I cannot
> imagine trying to put those two documents on the web as html documents.

I'm with TS on this one. When you need to control what the printed page
looks like, PDF is the way to go. In this application, the printed page
is more important than the on-screen display.

> PDF seems great for this sort of thing. However, as I explained, i am
> getting INCONSISTENT conversion. Sometimes the PDF documents have
> embedded Vectors and sometimes they have embedded bitmaps.

How do you know that, other than looking at them in 6400% magnification?
How do you know that what you are looking at is really a vector drawing
and not a very high resolution bitmap?

TS Moderator

2006-09-24, 6:58 pm

Scott Bryce wrote:
> How do you know that, other than looking at them in 6400% magnification?
> How do you know that what you are looking at is really a vector drawing
> and not a very high resolution bitmap?

File sizes are two small to support bit mapping that good at 6400%
resolution.

Scott Bryce

2006-09-24, 6:58 pm

TS Moderator wrote:

> File sizes are two small to support bit mapping that good at 6400%
> resolution.

You may be right. Greyscale line drawings compress very well, but
probably not that well.

The page content in the file is encoded. Otherwise I could tell you if
it was using any vector images. Each page has around 15 bitmapped images
in it, but I can't tell what those images are or if there are also
lines, curves, etc. My knowledge of the PDF spec is limited, and I don't
have tools to extract anything from the file.

It looks like you haven't had any success at comp.text.pdf yet.

TS Moderator

2006-09-24, 6:59 pm

Scott Bryce wrote:
> TS Moderator wrote:
>
>
> You may be right. Greyscale line drawings compress very well, but
> probably not that well.
>
> The page content in the file is encoded. Otherwise I could tell you if
> it was using any vector images. Each page has around 15 bitmapped images
> in it, but I can't tell what those images are or if there are also
> lines, curves, etc. My knowledge of the PDF spec is limited, and I don't
> have tools to extract anything from the file.
>
> It looks like you haven't had any success at comp.text.pdf yet.

Hey Scott,

I wrote an email to a bunch of the PDF conversion companies. Getting
some interesting answers, I will summarize and post when I have a
little more information. Thanks for your help.

- Sam -