AI绘图大师

About this Skill

Visual analysis and text-to-image prompt engineering using the SOCAWT framework. Supports two modes: (1) direct prompt creation from user descriptions, and (2) image analysis with prompt extraction via multimodal LLM. Detects image type first, then applies adaptive analysis depth and the appropriate prompt template. All prompts are written in natural language sentences, suitable for modern models like Jimeng, GPT-Image-2, and Nano Banana.