What problem does it solve? Writing effective prompts for AI image generators (NanoBanana, Gemini, DALL-E, Midjourney) to produce publication-quality academic figures is difficult: prompts are often too vague, lack color discipline, and omit structural details, resulting in figures unsuitable for top conferences like CVPR or NeurIPS. ## Core Features & Use Cases - Structured Prompt Templates: Generates prompts in four layers (global description, section-by-section detail, global annotations, style specifications) for framework diagrams, network architectures, module details, comparison/ablation figures, and data pattern grids. - Eight Preset Color Palettes: Offers curated schemes (Okabe-Ito, Blue monochrome, Teal+Amber, Navy+Coral, etc.) with exact hex values, plus links to external color tools like Coolors and ColorBrewer. - Paper-Aware Workflow: Reads LaTeX, Word, or PDF sources to extract model architecture, data flow, and math notation before generating prompts, with a quality checklist enforcing information density and grayscale readability. - Use Case: A researcher finishing a CVPR paper uploads the LaTeX source and asks for a framework diagram prompt; the skill analyzes the pipeline, confirms a color scheme, and outputs a dense English prompt ready for Midjourney. ## Quick Start Read my attached paper and generate a detailed English prompt for an overall framework diagram using the Okabe-Ito color scheme.
