Art as a probability distribution, not a point
https://gyazo.com/b7e792019255d99e0bb43eeb1ded29b5
Recognize in a set of multiple works by a single artist
This can be interpreted as approximating the distribution shape with a finite set
Then why not model the distribution itself with a better means of representation?
For example, giving a certain prompt P for Stable Diffusion means that there is a distribution of pictures defined by the multiplication of P and a seed s that follows a uniform distribution and sampled from it.
The prompt is automatically generated with a fixed seed, which is a multiplication of p sampled from the distribution of prompts and a fixed seed S.
2022-10-12
Cézanne began to question the techniques of the Impressionists when he was about 40 years old. He thought, "They only depicted the 'light' of the moment and failed to depict the essence of the trees and leaves, which is the most important thing.
For example, the "tree" that the Impressionists were painting was "a zelkova tree seen at 15:37:54 on August 13, 1850, at 5, Barbizon Village, Paris. But Cézanne wanted to paint "a very common and very permanent zelkova tree. A zelkova tree that can be recognized by anyone, anytime. I thought, "That figure is the essence of the motif.
He says, "Monet (a leading impressionist painter) has a humongous eye. But it is only 'one eye'". In other words, he thought, "I should paint motifs that can be seen and sympathized with through the eyes of everyone in the world.
The means employed were different, well there was no computer and
---
This page is auto-translated from /nishio/点ではなく確率分布としてのアート using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I'm very happy to spread my thought to non-Japanese readers.