BLIP-Diffusion is a multi-modal controllable theme-driven image generation model. It takes a theme image and a text prompt as input to generate a theme image based on the text prompt. Compared to existing models, BLIP-Diffusion offers zero-shot theme-driven generation and efficient custom theme fine-tuning. It can also be flexibly combined with other techniques (such as ControlNet and prompt-to-prompt) to achieve novel theme-driven generation and editing applications.