[Printers] Remedying the lack-of-data problem

[Seth David Schoen]

As many people are aware, EFF has thousands of test pages printed on
over 100 different printer models here in our office, thanks to
submissions from all around the world.  They're still coming in,
especially after the recent press attention.  (Our original deadline
for submissions will expire at the beginning of next week; we may put
together a new test sheet to reduce printing artifacts and to ask
people explicitly for the date and time that they made a given print.)

In the meantime, several people have pointed out correctly that an
ordinary scanner can pick up the dots when a scan is made at high
resolution.  (We don't need to modify the scanner by adding blue light,
as Bunnie did; the built-in blue light in a scanner is strong enough.)
So I'm going to try to get ahold of some volunteers to come and scan at
least a few hundred test pages, so that we can increase the amount of
raw data that's available to the community.

Unfortunately, while most of those do have serial number information
available, they generally don't have date and time information, so the
amount of reverse engineering we can do for printers that do print
date and time will be limited.  But I hope we can get a lot of actual
dot pattern data out there quickly.

I'm also excited about Ralf Muschall's FFT autocorrelation program,
and I plan to talk this over with some colleagues who are familiar with
signal processing.  Ideally, we should be able to convert a directory
full of TIFF files into a directory full of text files containing the
detected matrix size for each model and a sample repetition of the
matrix (and perhaps automatically pointing out some interesting
properties like parity).  Unfortunately, there are some tracking codes
that are not aligned to a rectangular grid (notably Canon's, which
sticks out like a sore thumb due to the vastly greater number of dots
and their more chaotic appearance -- although they _are_ still periodic,
which will ultimately help us quite a bit).  They may, as some people
have pointed out, code data based on the position of the dots relative
to some grid or basic pattern, rather than on the presence or absence
of dots.

I'm most excited to attack the Dell dots, because Dell color lasers
are about to be in a huge number of homes as a result of their low
price.  But I believe we only have three or four Dell samples so far.
Maybe the next batch of samples will increase that number a little.

-- 
Seth Schoen
Staff Technologist                                schoen at eff.org
Electronic Frontier Foundation                    http://www.eff.org/
454 Shotwell Street, San Francisco, CA  94110     1 415 436 9333 x107