This project leverages the programming capabilities of Large Language Models (LLM) to help users automatically complete the prompt words for mind mapping. It can generate high-quality images based on brief user-provided prompt words and supports features like localized image modification, such as turning a dragon into a dinosaur within an image, significantly reducing the barrier to creating satisfying images without the need for complex prompt words.