ajinnaaa/extract-visual-style-2
Data science & MLThis page was translated automatically. Technical details may differ from the original. View the original
Extract project-level visual systems, visual bibles, and Codex Skills for prompts from reference materials
Overview
A project-level visual style extraction skill for Codex. It can extract transferable colors, lighting, space, composition, materials, people, and motion systems from images, animations, movies, TV, advertisements, or video references, and organize them into executable project visual bibles and generated prompts. This project is an original personal skill and is not an official product of OpenAI, Codex, or GitHub. - Distinguish content from reference materials, transferable styles, production locks, and scene-specific content - Extract stable cores, conditional variants, and scene-specific rules - Analyze live-action images, animations, illustrations, and stylized CG - Output item-level visual bibles, style parent prompts, and result acceptance criteria - Clearly distinguish observations, inferences, suggestions, and unknowns when evidence is insufficient These six files together form a complete skill. Please do not just copy SKILL.md, or you will lack media analysis rules and project-level visual bible patterns. 1. Download or clone this repository. 2. Copy the entire extract-visual-style-2 folder to: - By default, only analysis and prompt delivery are provided; Unless explicitly requested by the user in the same request, images or videos will not be generated. - Specific cameras, lenses, film, LUTs, software, or production processes will not be asserted solely based on the appearance of the image. - When using third-party film, photography, or design materials, please confirm copyright, portrait rights, and platform rules yourself. - This repository does not contain sample films, characters, scene images, or third-party fonts. This project uses the MIT License. Use, copying, modifying, and republishing are permitted, but the original copyright notice and license text must be retained.
Overview (machine translated)
A project-level visual style extraction skill for Codex. It can extract transferable colors, lighting, space, composition, materials, people, and motion systems from images, animations, movies, TV, advertisements, or video references, and organize them into executable project visual bibles and generated prompts.
This project is an original personal skill and is not an official product of OpenAI, Codex, or GitHub.
What can be done?
- Distinguish content in reference materials, transferable styles, create locks, and scene-specific content
- Extract stable cores, conditional variants, and scenario-specific rules
- Analyze live-action images, animation, illustration, and stylized CG
- Output project-level visual bibles, style parent prompts, and result acceptance criteria
- Clearly distinguish between observations, inferences, proposals, and unknowns when evidence is insufficient
Directory Structure
These six files together form a complete skill. Please do not just copy SKILL.md, or you will lack media analysis rules and project-level visual bible patterns.
Installation
-
Download or clone this repository.
-
Copy the entire
extract-visual-style-2folder to: -
Reopen Codex or create a new task.
Usage
It can be triggered in the following ways:
Or simply enter:
Project-level test example:
Use boundaries
- By default, only analysis and prompt delivery are provided; Unless explicitly requested by the user in the same request, images or videos will not be generated.
- Do not assert specific cameras, lenses, film, LUTs, software, or production processes based solely on the appearance of the image.
- When using third-party films, photography, or design materials, please confirm copyright, portrait rights, and platform rules yourself.
- This repository does not contain sample videos, characters, scene images, or third-party fonts.
License Agreement
This project uses the MIT License. Use, copying, modifying, and republishing are permitted, but the original copyright notice and license text must be retained.
This page was translated automatically. Technical details may differ from the original.
Recommended Tools
Try a different keyword or remove a filter.
Install
npx skillfish add ajinnaaa/extract-visual-style-2