Save and Share:
The first public release of Stable Diffusion, a neural network that generates images from text descriptions.
Unlike the well-known DALL-E 2 and Midjourney neural networks, Stable Diffusion can be run on your own computer with a consumer graphics card.
The neural network’s open source code and launch instructions are located in a repository on GitHub.
The neural network has been deeply optimized for release: Stable Diffusion currently requires about 6.9 gigabytes of video memory, and optimization work will continue.
Running Stable Diffusion is recommended on Nvidia graphics cards. Special versions of the neural network optimized for AMD graphics cards and Apple’s M1 and M2 chips are also planned for the future.