What is personal-ip-scenes?

This repository is a "skill" for AI coding assistants like Claude or Codex. A skill is a set of instructions the assistant can follow to perform a specific creative task. In this case, the task is generating illustration images for Chinese-language articles, featuring a personal IP character. The skill works in two stages. First, you give the AI a real photo of yourself. The AI turns that photo into a stable hand-drawn character, producing two reference images. One image locks down your identity details, like facial features and clothing. The other image shows your character performing common actions, such as working at a desk or writing notes. This establishes your personal IP, which is a cartoon-like version of you that stays consistent across future images. Once your character is initialized, you can use it to generate illustrations for your articles. You ask the AI to create images based on your article's content, and it produces pictures featuring your personal IP character. The style is meant to look like slightly odd hand-drawn sketches on white paper, with lots of blank space and a few handwritten Chinese notes. The character must be doing something central to the image, not just standing in the corner. The project draws inspiration from a similar project called Xiaohei Skills, which used the same workflow of a fixed character, white backgrounds, and handwritten annotations. However, this repository does not include a default character. Every user must initialize their own personal IP from their own photo. The repository includes a profile file to record stable identity traits, clothing hints, and things to avoid, so the AI can maintain consistency. The README recommends using a specific image model, GPT Image 2, for the best results. It notes that this model provides more stable character consistency and the desired hand-drawn style. The repository is structured as a folder you copy into your AI assistant's skills directory, and it includes example images and an MIT license.