The Generative Art Community experiences a huge high with the release of an increasingly popular VQGAN+CLIP notebook, which generates unique art given an input description.
The description is allowed to contain any word describing anything, including styles descriptions, like “by Greg Rutkowski”. Here’s an example:
I’ve been playing with the provided notebook and I find that the pre-trained neural network and OpenAI’s CLIP algorithm, manages to produce incredible inspiring and intriguing results.
It’s as if the generated art calls to the mystery behind the words, showing us how the “summary of the visual internet” can be re-purposed to generate more content.
The art pieces in this article are one-of-a-kind. Click on the images to view the entered description.
More resources if you’d like to create these yourself:
Also published on Medium.