Stable Diffusion and crazy animations
Neural networks that create images based on text or voice descriptions have only recently begun to appear. But they have already gone from fanciful abstractions to almost photographic images that are generated in minutes.
The Stable Diffusion neural network, which appeared in mid-August 2022, generates images of 512×512 pixels in just a few seconds under all conditions. This system is one of the biggest competitors for Dalle-2.
Xander Steenbrugge, a programmer and artist from Belgium, used Stable Diffusion to show how to play with incoming noise and vectors to get crazy animations and turn an elephant into an octopus.
Turns out #stablediffusion can do really awesome interpolations between text prompts if you fix the initialization noise and slerp between the prompt conditioning vectors: pic.twitter.com/lWOoETYVZ3
— Xander Steenbrugge (@xsteenbrugge) August 7, 2022
Stable Diffusion is currently in closed beta, but promises to share the code soon. Then Disco Diffusion will come to life with a new architecture and dataset! And there will be a new burst of crazy animations!