Re: [Groff] Embedding encapsulated PostScript page

[Tadziu Hoffmann]

> 1. Resolution
[scanning in integer multiples of 72 dpi]
> Is this assumption right?

No.  The basic axiom is that the *final* image to be printed is
at about device resolution -- let's make it a factor of two
better, just to be on the safe side.  So, if your printer's
resolution is r, the final (printed) image width is xp, and the
width of your original image (the one you're scanning) is xs,
the you need to scan at

  s = 2 * r * xp / xs

(More is not economical because the file size goes up without
visible improvement, significantly less is also bad because
you're not using your printer to its full capacity, and you
will begin to see the pixels of your scanned image.)

The above is valid for monochrome (black-and-white) images.
When printing grayscale images, it depends on whether your
printer can display shades of gray already at device resolution
(e.g., dye-sublimation printers or (most) computer displays),
or whether halftones are approximated by dithering or clustering
of monochrome pixels (laser printers, offset printers, most
inkjet printers).  In the first case, use the same formula as
above.  In the latter case, you can substitute an "effective
grayscale resolution" for r, namely

  r_eff = r_dev / sqrt(# of graylevels)

where r_dev is your printer's intrinsic monochrome resolution.
Postscript let's you specify custom halftoning screens to
control the tradeoff between effective resolution and
graylevels.

Printing in color is essentially the same as printing several
monochrome images on top of each other with different inks.


> 2.This one seem a little naive even to me, but I don't
> understand this: As Ted describes it the printer literally
> print on top of the eps when doing a "watermark". Now how is
> it possible for a printer to print white text on top of a dark
> blue eps (I guess that white is the same as printing noting
> for the printer).

Think of Postscript printing as a two-stage process.  In the
first step, the image is printed to an internal bitmap, using
opaque (i.e., non-transparent) inks.  Printing white over blue
simply means that a pixel that was blue will be turned white.
Only when the page image is fully completed will it be transferred
to paper in a second step.  In this step, printing a white pixel
really does mean "don't put any ink here" (assuming the paper
is white).


> 3. Netpbm lets you choose if you want some kind of compression,
> to make the information smaller. It is the same kind of
> compression that png is using. Would you use such a possibility
> or do you loose to much quality by that?

PNG compression is lossless, so it doesn't degrade the quality.
Incidentally, Postscript has a builtin filter (FlateDecode) to
decompress the actual data, but you need to handle the details of
the PNG file format yourself (in contrast to JPEG, where you can
throw a complete JFIF file at the DCTDecode filter).