This project uses gpt-3.5-turbo and stable-diffusion-xl-base-1.0 & stable-diffusion-xl-refiner-1.0 to generate children's bedtime stories with illustrations. These stories can then be read aloud for you while showing you the approriate image at each point of the story.
The gpt-3.5-turbo model is used to generate the story text and image prompts. These image prompts get fed into stable-diffusion-xl models to produce the illustrations. The stories can be played using a Raspberry Pi with a display, although this is not required. It utilizes OpenAI's API and Hugging Face's diffusers library.
Modifications to the original project include using stabilityai/stable-diffusion-xl-base-1.0 rather than DeepAI's API. The model is considerably slower than other options but produces better images.