I shall try to explain how watermarking works. There are two kinds of watermarks: - The classic watermark where you add a soft watermark image to your picture, you can do it easily with layers in Photoshop. This watermark is always visible. - The digital watermark, this is not visible, in case of a strong watermark you will see noise and distorsions in your image. This kind of watermarks is very much difficult to explain, I shall try to do my best. You have some function between two variables, for example I = f(t), where I is the intensity and t is the time. Now by a mathematical process you transform your function I = f(t) into another function I = g(w) where w is the frecuency. You are representing the same data in another form. Too complicated?, let try a practical example. The variable I is the intensity of a music, a sound is a wave whose intensity is changing through time and fast. You can record the sound wave with a mechanical recorder or view it in an osciloscope. But there is another way to analyze a music, you can analyze it speaking in bass, middle, treeble tones. Now we are speaking in sound frecuencies and not in time with the shape of the waves. Now you have two ways to describe a sound, in the temporal domain of intensity and time or in the frecuency domain with intensity and frecuency. The mathematical function that allows us to change one representation into another is the Fourrier transform. Now we can go back to watermarking. With images is more complicated because we have three variables, the intensity of the pixel (for simplifying only grayscales) and two coordinates x and y (or height and width), so our image is a function I = f(x,y). Now we do the same as with sound and apply another mathematical transformation that can be a two dimensional Fourrier transform and we obtain another representation of the same data of the image I = g(wx,wy). Both ways I = f(x,y) or I = g(wx,wy) are representing our image. Now we take the other reprentation and we change a little I = g(wx,wy) + watermark(wx,wy) and taking the inverse mathematical process (inverse Fourrier transform) and we return to the normal representation of our image I' = f(x,y) this time the image is watermarked. Is something equivalent to adding a treeble note to the music, but our ears are sensible to frecuencies and if we listen carefuly we can hear the watermark. Our eyes are only sensible to spatial information and not to frecuency, so we are not able to see the watermark. Don't kill me, I did my best....

Stupidity also evolves!