1994-06-17 - Re: Bart Nagel in Mondo

Header Data

From: tcmay@netcom.com (Timothy C. May)
To: whitaker@dpair.csd.sgi.com (Russell Whitaker)
Message Hash: fa685e2f35b8b3bee8a7c0e5ef8967ca81a8a18a8588de8a650c85ab27e22a4d
Message ID: <199406171954.MAA07081@netcom5.netcom.com>
Reply To: <9406171227.ZM4384@dpair.csd.sgi.com>
UTC Datetime: 1994-06-17 19:55:27 UTC
Raw Date: Fri, 17 Jun 94 12:55:27 PDT

Raw message

From: tcmay@netcom.com (Timothy C. May)
Date: Fri, 17 Jun 94 12:55:27 PDT
To: whitaker@dpair.csd.sgi.com (Russell Whitaker)
Subject: Re: Bart Nagel in Mondo
In-Reply-To: <9406171227.ZM4384@dpair.csd.sgi.com>
Message-ID: <199406171954.MAA07081@netcom5.netcom.com>
MIME-Version: 1.0
Content-Type: text/plain


Russell Whitaker writes:

> [following up on the above message of mine earlier....]
> 
> Romana has asked me to pass along the following:
> 
>   In this month's issue of Photo/Electric Imaging,  Stego is featured as an
>   ideal product for "watermarking" digital images.  A commerical photographer
>   pointed out the article to me.   This magazine is available at major camera
>   stores such as Ewert's in San Jose - I haven't  dropped by to pick up my
>   own copy yet. I was polite but icy to Mondo folks when they inquired about
>   Stego, sent them the same info that I send everyone else, but pointed out the
>   exceedingly poor research they had done for _Mondo's Guide To The New Edge_,
>     and expressed a hope that my product would not be similarly treated.
> 
> [End of excerpted forward to cypherpunks by Romana Machado (romana@apple.com)]

I've been writing about the LSB method of steganography in digital
images and sound files since 1988 (in sci.crypt, elsewhere), but I
find this "watermarking" idea extremely implausible:

- the LSB method works for digital images, but is easily defeated by
twiddling the low order bits (essentially overwriting the LSBs...if
the first set is visually undetectable, so will later overwrites).

- the most common method by far of "appropriating" images is via
scanners. These scanners will not generally pick up the LSBs, or even
the second-least significant bits. And if they do, some slight
blurring or filtering would remove them. And so on.

(This is what clued me in 3.7 seconds into reading Bart Nagel's piece
in "Mondo." His mention of "cypherpunks" was a nice touch, too.)

"Watermarking" of analog images is impractical, at least using any
variants of the LSB method I can think of. And digital images can
similarly be filtered/convolved so as smear or blur any such
signatures beyond recognition. In the final analysis, any file that
can be viewed by the eye, or heard, can be stolen.

(Higher-level issues of recognizing images may be different. For
example, the Pebble Beach Golf Course has _claimed_ that all
photographic images of the famous "lone cypress" belong to it...and it
tries to collect royalies from anyone who publishes a picture of it!)

--Tim May


-- 
..........................................................................
Timothy C. May         | Crypto Anarchy: encryption, digital money,  
tcmay@netcom.com       | anonymous networks, digital pseudonyms, zero
408-688-5409           | knowledge, reputations, information markets, 
W.A.S.T.E.: Aptos, CA  | black markets, collapse of governments.
Higher Power: 2^859433 | Public Key: PGP and MailSafe available.
"National borders are just speed bumps on the information superhighway."




Thread