Neural image colorization algorithms hallucinate color in grayscale pictures. Yet, the task is impossible. They can only produce outputs that we perceive as plausible but potentially wild guesses of reality: “plausible perceptibility is not reality”!
Consider a picture of Satan’s bolete (left). The red trunk with a white hat looks bad and dangerous. Now, take a grayscale version of it (middle) and colorize it with your favorite colorizing API (i.e., a neural network algorithm). We end up with a delicious bolete (right).
Style transfer algorithms harmonize two art expressions that were not meant to meet.
To dialogue, the artist duo Tina&Charly either paint or dance.
While motion is a fundamental aspect of their artistic approach, their paintings are simply not moving. Conversely, their video record all their dances but do not retain their painting aesthetic.
Learn how previously discarded style transfer instabilities can create psychedelic videos
The first frame of the video, a watercolor, is progressively stirred into a plethora of curvy and colorful patches. It then metamorphoses into a purplish phantasmal coral reef that is itself slowly submerged by an angry puce ocean. The water then calms down as the coral reef disappears and ends up perfectly still. How is this psychedelic animation related to style transfer methods?
Neural style transfers are rendering techniques — for images mostly — that seek to stylize…
I recently found some old yellowed watercolors by my grandfather.
To revive these, I ran neural style transfer methods. I considered my Brittany watercolors as style images and my summer hiking pics, in the San Juan National Forest, Colorado, as content images.
It was a winter forest in summer, most of the trees were dead, killed by some bug, an ominous warning…