
youxichuhai/character-animation-skill
Developer toolscharacter-animation-skill
Overview
A local Agent skill for turning a user-supplied or ImageGen-created character reference image into a short action video, PNG animation frames, and a sprite sheet. The included provider is configured through a generic image-to-video API adapter. API keys are read from environment variables only. Requirements: - Python 3.10+ - FFmpeg available on PATH - Pillow - PyYAML - requests Set your video API key in the shell environment. Do not paste secrets into source code or commit .env. - an existing character image supplied by the user; or - a new character reference generated with ImageGen inside Codex. For generated references, create a single full-body 2D character on a simple or transparent background, then use that image as the --image input. Read SKILL.md for the Agent workflow and first-run setup behavior. This repository includes one authorized example:
README
Character Animation Skill
English | 简体中文
A local Agent skill for turning a user-supplied or ImageGen-created character reference image into a short action video, PNG animation frames, and a sprite sheet.
The included provider is configured through a generic image-to-video API adapter. API keys are read from environment variables only.
Quick start
Requirements:
- Python 3.10+
- FFmpeg available on PATH
- Pillow
- PyYAML
- requests
Install Python dependencies:
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
Copy configuration:
cp config.example.yaml config.yaml
cp .env.example .env
Set your video API key in the shell environment. Do not paste secrets into source code or commit .env.
export VIDEO_API_KEY="your-api-key"
You can start from either:
- an existing character image supplied by the user; or
- a new character reference generated with ImageGen inside Codex.
For generated references, create a single full-body 2D character on a simple or transparent background, then use that image as the --image input.
The provider must implement:
generate(image_path, prompt, duration, output_path) -> local_video_path
Once configured:
python scripts/generate_video.py --image character.png --action "walk" --duration 4 --loop --output output/source.mp4
python scripts/extract_frames.py --video output/source.mp4 --output-dir output/raw_frames --fps 8
python scripts/align_frames.py --input-dir output/raw_frames --output-dir output/frames
python scripts/build_spritesheet.py --input-dir output/frames --output output/spritesheet.png
Read SKILL.md for the Agent workflow and first-run setup behavior.
Example
This repository includes one authorized example:
examples/chara.png: source character sheet.examples/reference_frame.png: cropped first frame used for image-to-video.examples/source_4s.mp4: generated 4-second source video.examples/spritesheet_8frames.png: compact 8-frame transparent sprite sheet.examples/walk_cycle_8frames_preview.gif: animated preview.
Generated working files should stay under output/, which is ignored by Git.
Recommended Tools
Try a different keyword or remove a filter.
Install
npx skillfish add youxichuhai/character-animation-skill