I only know a little about steganography in the grand scheme of things, but I have heard from people who generally turn out to be right, that there is no such thing as strong steganography. If an image or other data is analyzed, it is possible to determine that it has something hidden in it. This is generally done with some statistical techniques looking for the presence of certain levels of randomness, since steganographically hiding data in a file always increases its randomness. It seems to me that maybe very small data transfers would be extremely difficult to detect though, I am kind of skeptic about adding a single additional bit of randomness being enough to identify stegos presence. This technique could also be countered by increasing the randomness in many images, but since most people are not doing this, the people attacking stego seem to be able to get almost non existent false negatives and few false positives, and to be able to fairly quickly defeat everything that is attempted. I believe using unknown and/or more sophisticated algorithms is helpful in slowing them down, but not in defeating their efforts. This doesn't mean that stego is not useful to avoid arousing suspicion in the first place, and in a way that is what it is intended for. But in another sense the goal of stego is to hide data in other data and hide the fact that this is happening, and in this area it seems to fail. Sorry I really know very little about stego, but it doesn't seem to be a favored technique by people who really know their shit.