--On Thursday, March 06, 2003 2:55 PM -0600 "David W. Tamkin"
<[EMAIL PROTECTED]> wrote:
> Tom Neff gave an example of non-HTML graphics:
>
> : There is also this method:
> :
> : JxIA0KDQo8aHRtb
> : 9uXCBjb2xvXj0iI
> : RybX5nPkRXTidUI
> : MgSUXgSEXTVE9SW
> : NlbnRXcXI+PGZvb
> : wgc2FuXy1zZXJpZ
> : JhdGVXbXJ0Z2FnZ
> : 1heSXiZSXhcyBsb
> : 9vcXBjcmVXaXQgb
> : BnXyB1cCEgXC9mb
> : NpemU9IjIiIGZhY
> :
> : What letter do you see in the above block?
>
> I saw nothing at first. Then I switched to a fixed-pitch font and still saw
> nothing. A second fixed-pitch font, still nothing. There's the catch: to
> make sure the reader has the right view, you have to sacrifice plain text
> and instead use some markup to specify a display font.
>
> Finally I traced all the capital X's and saw that they form a capital X, but
> if my earlier examples are too involved for the typical human reader, this
> one is scarcely any better. Few people would bother, and all told I think
> this method is no better than those I tossed out earlier.
In my own lame defense :) I just hacked that together - I actually stole a
rectangle of random Base64 from some Zip file someone sent me and used it as
the palimpsest. I agree the results are pretty bad. I had seen other
examples of this kind of multiline pattern recognition that were a lot easier
- but they used dots and other low-profile characters for the "background."
I thought that might be a bit easy to reverse engineer so I tried the random
characters. See if you can read this:
.,.._...,_,..,,
_,X..,,_,,X,_,.
.,,X_,.,.X.,,..
.,..X,..X..._..
.,,,.X,X.+...,,
,,,_.,X,_,.;.,.
.,,..X,X._._.,.
_,,.X,..X,,,.,,
_,,X.,,,.X,;.,,
.,X,._,..,X._,,
.,,,._.,.,...,.
