JoyCaption is a Visual Language Model designed to generate captions for images. It is built from the ground up as a free, open, and uncensored model to serve the AI art community. The model processes images and outputs detailed textual descriptions, which can be used as training data annotations for Diffusion models. Training scripts and methodology details are released openly with the weights.