graphics becoming corrupt

[rickk -at- slmd -dot- com]

Has anyone else experienced the following odd sort of corruption in
imported graphics? Does anyone know how to fix it? This problem started
when we upgraded from Framemaker 5.5.3 to 5.5.6 (and it has continued with
our more recent upgrade to 6.0), but we don't know if that is a cause or a
coincidence. The corruption appears in imported graphics (both "by copy"
and "by reference") when we open a Framemaker file. I'll describe it first
and then add facts about the circumstances below.

The corruption has two basic forms:
(1) A horizontal row of pixels, say 10 to 50 pixels thick, running the
length of the graphic will be shifted to the left or right about a quarter
inch, so it is not aligned with the rest of the graphic. It looks on screen
(and in print out) as though someone used sissors to cut a thin horizontal
strip out of the graphic and slid it leftward (or rightward) a quarter of
an inch.
(2) Two squares of pixels, each consisting of about 1/6th of the surface
area of the graphic, have swapped places. So, for example, what should be
the lower left corner of a screen shot appears in the lower right of the
screen shot, and what should be the lower right is in the lower left. All
the pixels are there, but two roughly square-shaped chunks of them have
swapped places.

Here are some facts that may or may not be relevant:
A. It appears on screen and in printout the same way. (I *think* that means
that nothing about the screen settings or display drivers could be the
cause, because those wouldn't affect the printout, but I'm not at all sure
of this.)
B. It happens on files (and graphics) that have been just fine for years
through successive revisions and publications (and many versions of
Framemaker), as well as on new files and/or new graphics. (I *think* that
means there's nothing intrinsically wrong with the graphics.)
C. Sometimes simply closing the Framemaker file and reopening it fixes the
problem. (But if the autosave goes off before we do this, the corruption in
the graphics becomes "permanent" until we find and copy or re-import an
uncorrupted version of the graphic.)
D. Sometimes opening the *.backup copy of the file shows that the graphic
is not corrupted in that file and we can copy/paste (but usually the
*.backup file has the same corruption).
E. Fixing it once usually isn't enough, it'll happen again the next time
the file is opened.
F. It is less likely to happen if the file is copied to a local drive and
opened from there as opposed to opening it in its home directory on a
server, but moving the file to a local drive only works about half the
time.
G. It happens on both "import by reference" and "import by copy" graphics.
H. It doesn't appear to correlate with graphic format. We've seen it happen
on *.png, *.gif, *.tif, and *.wmf graphics among others.
I. It doesn't appear to correlate with the size of the graphic, either in
bytes or surface area.
J. If it happens on one graphic in a file, it will almost always happen on
several in that same file, but it never seems to happen on more than 1/3 of
the graphics in a given file.
K. When it happens a second (or third, etc.) time to the same file (see
item E above), it is mostly, but not completely, the same subset of
graphics that are corrupt. A few graphics that were OK the last time the
file was opened are now corrupt and a few that had to be fixed before are
OK now.
L. It happens on more than one computer in our department.
M. We have several postscript drivers, including Adobe 5.1.2, and switching
among them has no effect.
N. We are on Windows NT, with service pack 5. We have 64 meg of RAM in our
machines.
O. None of us can recall it happening before we upgraded from Frame 5.5.3
to Frame 5.5.6 last fall, but that might be coincidence. (At around that
same time we installed NT's service pack 5.)

Have any of you experienced this? Any ideas on a fix?