Artificial intelligence imaging is in full swing. There are many online tools to create drawings, logos, art of various kinds, even music. In this post I share a search engine for images belonging to Stable Diffusion called Lexica.
Stable diffusion
Stable Diffusion is a machine learning model developed by Stability AI to easily generate high-quality digital images from worded descriptions. The model that is created uses a seed, a starting value, that generates the model. This seed allows us to obtain the same result if we use just the same seed to generate another model.
This example is what it would be like if we lookedwith a Halloween and clowns’ theme.
Available code
The Stable diffusion code is available here and the model here. Stable Diffusion is a text-to-picture model that will allow billions of people to create art in seconds through artificial intelligence.
The model uses a widely used latent diffusion model combined with knowledge of generative AI developer Katherine Crowson’s conditional diffusion models, Dall-E 2, Open AI, Google Brain Image and many others. AI media generation is a cooperative field and is going to go a long way in the coming years.
Project documentation
Lexica has documentation that can be found in https://lexica.art/docs. This link allows us to use the Lexica API through a few simple lines of code. The references also share a Google Colab book that also allows you to create models.
In this post show how to use Stable Diffusion with the 🧨 Diffusers library, and explain how the model works and finally dive a bit deeper into how diffusers allows one to customize the image generation pipeline.
