This was an experiment to see how much text I could fit into a steganographic image. This image contains the complete text of Melville's two novels, Moby Dick and Typee.
To decode it, you will need access to the original image, which I have stored here (the original website I found the image on is no longer up).
Each pixel stores 12 bits of hidden information, the data was compressed into a zip file and then used to modify the pixel values.
Interestingly, this method of steganography produces some significant compression savings:
Zipped Text: 693k
Original PNG file: 998k
if I simply zip these two files together, the result is 1,691k. But by using
steganography, the output PNG file is only 1,191k - a savings of 500k.
I assume the savings is related to the fact that you can't recover the text without access to the original image - there is not enough information in the output file to recover both input files.
More stuff by jbum: