yang0/handraw-style

What it solves

This project addresses the difficulty of describing specific artistic styles and layout compositions in AI image generation. It eliminates the need to memorize complex terminology by providing a numbered system for styles, layouts, and colors, ensuring consistent and high-quality visual outputs across different AI tools.

How it works

The system operates as a "Skill" (compatible with Codex) that translates user themes into precise, bilingual prompts. Users can interact with it in several ways:

• Numbered Selection: Users pick a style number (from 327 options), a layout number (from 165 options), and a theme color (from 36 options) to generate a prompt.
• Intelligent Recommendation: The AI automatically matches the best style and color based on the theme's mood and atmosphere.
• Specialized Modes:
  • Poster Design: Generates structured prompts for high-impact visuals with integrated text.
  • Cover Design: Analyzes articles or video subtitles to create visual metaphors and eye-catching covers.
  • Batch Illustration: Plans visual anchor points in long-form text and generates a unified set of images to be inserted back into Markdown.
  • Photography Planning: Creates professional-grade photography schemes for couples, weddings, and portraits.

Who it’s for

• Content Creators: For social media (Xiaohongshu, WeChat) covers and posters.
• Knowledge Bloggers: For creating infographics, process diagrams, and educational long-form images.
• Storytellers: For comic storyboards and picture book illustrations.
• Designers: For rapid prototyping of hand-drawn sketches and creative posters.
• Photography Enthusiasts: For planning professional-style portrait shoots.

Highlights

• Extensive Library: 327 hand-drawn styles, 165 layout types (social cards, infographics, comics, IP design, e-commerce), and 36 classic theme colors.
• Dual-Mode Output: Supports "Pure Image" mode for illustrations and "Graphic-Text" mode where text is integrated into the composition.
• Model Adaptation: Uses explicit keyword activation for optimized models (like gpt-image-2) and reference image fallbacks for others (Midjourney, Flux, Stable Diffusion).
• Automated Workflow: Includes features for automatic article summary extraction, visual metaphor design, and automated Markdown image backfilling.