So someone in this big ol' planet found my little blog entry about the Hanso Foundation image and linked to it on Technorati.   I followed a thread about this and found that at least a few people thought I was crazy and had no clue what I was talking about.  Well, yes and no.

I processed the image I found exactly as I said – splitting it into it's HSV component images using The GIMP .  The non-random areas are suspect, but that's not proof there is steganographic material in there.  I need to run it through real steganographic detection algorithms to see if the steg is there.  And, at least according to one post, I need to run the correct image through that process.  I used the combined JPEG found at The Lost Experience web site.  Whether or not that's the correct image to use, I don't know  (apparently the image to process is on the Hole2 web site instead, which is a GIF image).  But I do need to run real steganographic algorithms on that image and not just do the HSV separation.  HSV separation just helped me identify if there were suspect areas in the image that don't show up as easily under normal RGB mode.  I tried various other separations too, but HSV had the most obvious results on the JPEG.

Well, that's what I'm going to do.  You see, I'm employed on a project to work on a distributed steganographic processing system.  Part of that project is a tool filled with steganographic detection algorithms.  I didn't write that part (I'm working on the grid stuff), but the guy who did is going to run the image through it and see what falls out.  I'll post more when I get some results.

What is funny is how some of these people (the posters in that thread) get so bent out of shape about these things.  I just did it for fun.  Relax folks.  It's just a show (albeit a really good one, I must admit).

Okay, so we finally ran the appropriate GIF file (not the JPEG I ran through the HSV splitter) through some of our steganalysis routines at work.  At the moment, none of the algorithms finds any hidden messages.  However, the fellow doing the tests says that doesn't prove anything yet.  He's got a ton of other detectors yet to add to his toolset and will keep looking at the image over time.

For now, however, we've got nothing hidden in the image.

If I can remember, I'll get a list of algorithms we ran on the image.  Most are publicly available detector algorithms.