Re: BlueSky fonts and subfont (and FixFont)

["Berthold K.P. Horn" <bkph@ai.mit.edu>]

   I wrote:

   ... and Berthold replied:

   >> The task is converting some legacy PostScript files (without the
   >> original DVI files that created them) from using bitmaps to using
   >> PostScript fonts, so neither dvips 5.66a nor dvipsone's partial font
   >> loading are likely to help.

   I forgot to mention here that I'm using the marvelous `FixFont' utility
   for this task.

Could you amplify on that?

   > How about the following: rip out the bitmap fonts.  Insert references
   > to the outline fonts, but *do not include the fonts themselves* Then
   > run through Distiller.  Take the resulting PDF file read it into
   > Acrobat Reader and print to file.  if you have told Distiller where
   > the fonts are and you have changes it settings to always subset you
   > will have what you want.

   Yes, this possibility occured to me (except that I'd have included the
   PostScript fonts (unsubsetted) in the file to distill, since distiller
   is of on another machine somewhere). 

Important point, as you say to do it this way you want the fonts to
be `unsubsetted'  Otherwise Distiller thinks they are complete fonts.
In your case they would naturally be unsubsetted, but I have seen
people produce bad PDF by having partial fonts in the PS files 
(as opposed to none which is by far the best - or complete fonts).

   One remaining issue is that even with the BlueSky fonts, replacing the
   PK fonts with type1 fonts doesn't result in the exact same output. Some
   positioning is subtly changed (in long runs of text) -- it's especially
   noticiable in things like tables of contents where the numbers on the
   righthand side turns out wobbly. Anyone know of why this might happen
   and better how it might be fixed?

This is because DVIPS output is not resolution-independent.  It
uses rounded off advance widths for the characters.  Normally these
`rounding effects' tend to cancel out since they tend to be random.
But you can see them build up when the same letter is used repeatedly
or in a fixed width font, where all the `rounding effects' have the
same sign.  The advance widths in the Type 1 fonts are exact
(fractional) and hence do not match the PK fonts advance widths exactly.

You don't see this problem when you produce output from DVIPS directly
using Type 1 fonts, since it forces the same fixed resolution onto
the Type 1 fonts by redefining their metrics.

   Regards,  Melissa.

   P.S. Also, does anyone know whether `t1disasm | t1asm' performed on a
   font is in any way lossy? (I'd like to be sure that I've not damaged
   my BlueSky fonts in way by piping them through these tools).

I believe they should be lossfree since they do not touch the hinting
or the outlines or change the subroutine structure.  You will lose
any hidden comments buried in the encrypted part.

Berthold Horn