integrating GPG with deniable steganography

[Bernd Eckenfels]

On Tue, Mar 20, 2001 at 09:01:51PM +0100, Matthias Urlichs wrote:
> I don't need random bits. I just need something to hide behind. That
> something can be randomness (it doesn't have to be 100% random; 10%
> random is fine too, I just need more bits -- the end result might even
> show _more_ correlation than the input file), but it can also be
> something entirely different.

No, the problem is that if you analyse typical noice (scanner, microfone,
CCD) you have a very characteristic picture, it is even possible to actual
fingerprint the device which has geenrated the data. If you add white
cryptographic "real true random" bits, those will be visible in filters
unless you take very much care. simply changing the LSbit of every <random>
byte will only bring you into trouble.

> For instance, I might take an MP3 encoder or a JPEG compressor.

JPEG Compressors actually dont generate any noice..  So by adding noice bits
you might actually generate files which could NEVER be genertaed by a
Compressor.

Greetings
Bernd