Re: SUMMARY AND FOLLOWUP QUESTIONS: Re: Acrobat file sizes

[Gilda_Spitz -at- markham -dot- longview -dot- ca]

Thanks again to all who took the time to answer my questions, especially
Max Wyss of Switzerland, who is obviously knowledgeable enough to write a
book on the topic of graphic file formats!

Combining Max's advice with some other techwr-l advice, I've decided to go
with GIFs. I realize we'll have to go back and redo many of our screen
shots, but - what the heck, if you make a mistake, you have to fix it,
right? It has taken several days, and we're not quite finished, but I'm now
convinced we've finally made the right decision.

For those of you who are interested, here's a followup summary:

From Beryl Doane:

GIFs are 256 color, and JPEGs are 16.2 million color files. JPEGs are
intended for color photographs where you need the additional color depth.
In
most cases, GIFs work just fine, but it does depend on the resolution and
color depth that you need.

Most graphics programs can save JPEGs as GIFs. Try converting a few of your
most complex files to see how they look.

There are a few batch programs that will convert graphics from one format
to
another, but I can't remember their names. Check the archives--I know I've
seen email recently. PaintShop Pro used to have a batch converter, but
again
I cannot remember its name.

From Seth:

I am not sure how quick a fix this will be but I suspect the file size
problem is related to your conversion to jpg.  Even though the jpgs
themselves are smaller, if acrobat is embedding them into the final PDF
(which in fact is usually what is done) then it is expanding the jpgs into
a format it can display properly.  BMPs compress extremely well--that is,
because this is a very common format, the compression algorithms are quite
common and this is probably why acrobat handles bmps so well.   JPG on the
other hand has several different formats.

I am guessing you went to jpg for web related reasons. Unless you are doing
very intense photography, GIF format might be your best option.  256 colors
is usually plenty to render almost anything with reasonable resolution.
Other than that, you are left with managing more than one format of the
graphics.

There are plenty of conversions tools (Lviewpro is one) that permit you to
quickly convert many graphics from one format to another.    I suggest
sticking with the original artwork format (ie.  Adobe Illustrator or
Photoshop) and then use the graphic conversion tool to convert everything
on the fly to BMPs or GIFs and then reference those in the PDFed document.


From Max Wyss:

There seems to be some confusion between the use of the JPEG graphic file
format in the creation/editing phase of the source document, and the use of
the JPEG compression in the creation of the PDF in Distiller. These two
topics must be kept apart.

For the source document, the JPEG graphic format is suitable for
photographic images. These are images where the color of a given pixel is
not determined by some arbitrary value, and where it does not influence the
actual appearance of the image (which is the case for photographs). For
this, the JPEG format is excellent. The JPEG graphic format uses the JPEG
compression, an algorithm which is "lossy". That means, it discards image
information which is kind of below a certain threshold. It relies on the
next neighbours. However one effect is that it creates artefacts around a
sharp contrast.

For applications where the color of a given pixel is determined by some
arbitrary value, a format is needed which does not modify that color.
Screenshots are examples for such an application. A format for these
applications must not be "lossy", but "lossless". The most widely used
lossless formats are TIFF and GIF (and all the raw formats, such as BMP on
Windoze).

TIFF can represent 8 or even 16 bit color depth per channel (and we have
three or four, depending on the color space we are using). It can use a
compression, but does not have to. The compression algorithms are part of
the run-length-encoded family.

The GIF format can represent up to 256 colors, selected out of a 8
bit-per-channel RGB range. This is, of course the reason why it leads to a
considerably smaller file size, as it uses only 8 bits color information
per pixel instead of 24 or more. The GIF format is in addition to that
compressing. The result is an even smaller file.

The raw formats do not contain any compression settings, and lead to really
big files. Some allow to set the color depth, some don't. They may have
some reason to be when it comes to fast display on a slow computer, as
there are no decompression algorithms which must run.

So, this covers the options you have when making your source documents.

For screenshots and Framemaker, I got the best results with the TIFF
format. I did process the screenshots in Adobe Photoshop: first, setting
the final dimension in which it gets placed in the Frame document, and then
a color reduction to only the needed colors, then saving as TIFF.


Now, for the compression settings in Distiller for images:

As said, this is something totally different. In a properly working
Distiller, we have several compression settings besides "none": JPEG at
several compression levels, ZIP-4-bit and ZIP-8-bit.

The JPEG compression is, as already stated "lossy". The compression levels
determine how big the losses are, and how much the quality of the resulting
image is affected. A high compression leads to a low quality, but a small
file (or image object in PDF). A low compression level leads to better
quality images, but bigger image objects. The optimum setting must be
determined for every individual document, as it is a compromise between
file size and image quality. The JPEG compression should -- particularly at
high compression levels -- only be used if the document contains only
photographic style images.

ZIP-4-bit is essentially a lossless compression. However, before it is
applied, the number of colors is reduced, so that the colors per channel
can be represented with 4 bits (which means 16 equidistant levels per
channel). Then, the ZIP compression algorithm is applied. The result is a
very small image object.

ZIP-8-bit is the only full quality lossless compression in PDF. It uses the
ZIP compression algorithm, and does not reduce the number of colors. For a
document containing screenshots, this compression setting is the only one
which ensures a sufficient quality.

So, back to your two questions:

1. Well, it would have been better to use a lossless compression file
format for your screenshots. But as you have them already, you would have
to decide if the quality is sufficient, particularly concerning the
artefacts along lines. If the compression has been set too high, and the
results are not really convincing, you might consider redoing the
screenshots. Otherwise, stay with the JPEGs, but don't make things worse in
Distiller, and use the ZIP-8-bit compression setting.

2. There are some tools which do batch converisions. Debabelizer comes to
my mind, which is said to be the industry standard for batch conversions.
If you want to do some additional processing, you might also consider the
batch processing capabilities of Adobe Photoshop.

<snip>

The reason why I prefer TIFF is that it behaves a bit better in Frame on
Mac, compared to GIF, and that it seems to run a bit snappier (because
Frame does not need to decompress the GIF on the fly, but can simply load
in the TIFF). It also does not have the 256 color limitation of the GIF
format. Otherwise, if your screenshots are always below 256 colors, I don't
see much of a problem using GIF instead.

That the GIFs are smaller than the TIFFs is understandable, as the GIF
format is a compressed format by design, whereas TIFF may or may not be
compressed. As you have noticed, when you end up in the PDF, there is no
real difference in size.