Reading Response 4 - Maddie Schroeder

            I have always been interested in photography, so when I stared reading Cory Arcangel’s article, “On Compression”, I was drawn to understand the process that goes into compressing information, especially images. I have worked with JPEGs before, but never actually knew what it stood for, Joint Photographic Experts Group, and what it meant to save a picture as a JPEG.

            He did a really nice job of explaining the process of compressing information through either Lossy or Lossless. His example of Lossless, saying that it is all the same information just saved in a more efficient way is really interesting. Instead of having ‘a a a a a a a a a b a’, it is just, nine ‘a’s’, one ‘b’, and one ‘a’, so all the information is there, but just simpler. To further explain Lossless, he says that this is what a zip file does when it is opened, it takes what is compressed and then reads it as the original version. When reading this example I immediately thought that this is what was used for images because I did not think there was any way an image or video could be missing information to be recognizable. When he proceeded to say that “believe it or not, our eyes and ears are pretty crap, and we don’t usually notice missing bits here and there” (Arcangel 221), I was not too convinced. But with his example explaining how instead of the nine ‘a’s’, one ‘b’, and one ‘a’, a compressed image would just be 11 ‘a’s’ it started to make sense, that our mind fills in the blanks and we generalize the images that we see, so a blurry or pixelated image is still recognizable.

            Looking into the mathematics of DCT formula, it amazes me of how people even think of these things! The equations are so complex and when he describes the different cosine frequencies, it gets a little confusing. What I find amazing is the fact that the pixelated version of the picture is only “the first few low frequency basis functions of an 8 x 8 2D DCT” (227). It makes me wonder what I lot of high frequency basis functions of a DCT would look like. Either way, this article has been very helpful with explaining some common computer vocabulary that I never really took into account. It was interesting to read and makes me wonder what images will look like in the future once compression has become even more revised.