Project 1: Noise&Form

Post Reply
alanasg
Posts: 6
Joined: Thu Jan 06, 2011 1:57 pm

Project 1: Noise&Form

Post by alanasg » Mon Jan 17, 2011 10:27 pm

When first reading Harmon’s article, I was struck by the way in which he described some specific set of calculations or guidelines that a picture must have, when noise is added, in order to be recognizable by the human eye, "When the hoise is in the band adjacent to the signal frequencies, it obscures the picture more effectively than when it is at least two octaves removed from the picture frequencies, confirming that critical-band masking is the most important mechanism limiting the recognition of degraded or blurred images such as block portraits". I was also fascinated specifically by noise and blur, and how two seemingly messy, random capsules of information had a way of making something unrecognizable, and then recognizable once more. This was my jumping off point for this project, I wanted to start with an image and blur it/resize it enough to make it unrecognizable, and then slowly attempt to bring it back to recognition. However, I decided to take it one step further and continue adding more and more noise to see where I would end up (exactly how far away from the original I could get). I started to think about the way in which enough noise seemed to equalize images, and even though you may think you’re creating a completely unique, random image, someone else could add a ton of noise (and because the whole process is done by a computer), you could end up with the same image as someone else, potentially (even though there are millions and millions of options). I didn’t like this idea, or where my project guidelines were taking me. Thus, I decided to invent a set of rules for myself and thought that, perhaps within these rules I could find some sort of freedom in my image-making process that I would not have otherwise found.

The rules I created began as an idea to imply the same standards to multiple, completely different pictures, to see what similarities and differences they held. What I found just so happened to not be visually interesting, so I expanded on the rules a bit.
(below are those images I started with)

next I decided to apply the same exact processes (noise, gaussian blur, saturation, rgb levels, etc), but with different numbers for each image. And this is what I ended up with.

San Francisco
sanfran.jpg
  • Resize: 8 x 600 - bicubic
  • Resize: 800 x 600 - bicubic smooth
  • Noise: 254%, gaussian
  • Blur: 23.3 px, gaussian
  • Levels: RGB 102, 1.0, 183
  • Levels: RGB 30, 1.0, 255
  • Saturation: +56
Grocery Store
grocerystore.jpg
  • Resize: 80 x 600 - bicubic
  • Resize: 800 x 600 - bicubic smooth
  • Noise: 334.13%, gaussian
  • Blur: 9.2 px, gaussian
  • Levels: RGB 101, 1.0, 255
  • Levels: RGB 42, 1.0, 131
  • Saturation: +89
Harbor
sandiego.jpg
  • Resize: 2 x 600 - bicubic
  • Resize: 800 x 600 - bicubic smooth
  • Noise: 101.25%, gaussian
  • Blur: 56.6 px, gaussian
  • Levels: RGB 102, 1.0, 255
  • Levels: RGB 0, 1.0, 64
  • Saturation: +83
Subway
subway.jpg
  • Resize: 800 x 6 - bicubic
  • Resize: 800 x 600 - bicubic smooth
  • Noise: 49.5%, gaussian
  • Blur: 31.8 px, gaussian
  • Levels: RGB 62, 1.0, 255
  • Levels: RGB 21, 1.0, 210
  • Saturation: -16
Farm
farm.jpg
  • Resize: 8 x 60 - bicubic
  • Resize: 800 x 600 - bicubic smooth
  • Noise: 400%, gaussian
  • Blur: 250.0 px, gaussian
  • Levels: RGB 125, 1.0, 255
  • Levels: RGB 0, 1.0, 2
  • Saturation: +100

Post Reply