Using MCMC to manipulate images
Published:
A few years ago, I was interested in exploring some ideas around the content of images and vision, and in trying to think of ways of reconstructing images from smaller amounts of information. At this point those who are “technically inclined” may think of compression techniques or using wavelet bases, but I was interested in doing something a bit more artistic. In playing around with this, I ended up coming up with an idea based around sampling a small percentage of pixels in the image and drawing paths between them, which to me gives rise to some interesting looking images: