PrettyComics

Generate comic-style illustrations from text descriptions using Google Gemini. Supports reference images for consistent character design and art style across multiple panels.

How it works

Text description + optional reference image → Gemini → Comic-style PNG

1. Describe a scene ("a robot fixing a server that's on fire")
2. Optionally provide a reference image for style consistency
3. Choose a style preset (comic, manga, cyberpunk, retro, pixel)
4. Get a comic-style illustration back as PNG

Quick start

As Claude Code plugin

/plugin marketplace add Jadefalkner/jadefalkner-plugins
/plugin install prettycomics@jadefalkner-plugins

Then use /prettycomics:prettycomic to create illustrations.

Standalone

git clone https://github.com/Jadefalkner/prettycomics.git
cd prettycomics
python3 -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt

Create a .env file:

GEMINI_API_KEY=your-key-here

# Optional: Kie.ai as fallback
KIE_API_KEY=your-key-here

Generate an illustration:

# Simple comic
python generate.py "a robot debugging code at 3am, surrounded by coffee cups"

# With reference image for style consistency
python generate.py "the same robot now celebrating a successful deploy" --ref previous-comic.png

# Different style
python generate.py "hackers in a neon-lit server room" --style cyberpunk

Options

generate.py "<description>" [options]

--ref PATH              Reference image for style consistency
--style STYLE           Visual style preset (default: comic)
-o, --output PATH       Output path (default: output/comic.png)
--resolution 1K|2K|4K   Resolution (default: 2K)
--aspect-ratio RATIO    Aspect ratio (default: 16:9)
--backend kie|gemini    Override rendering backend

Styles

Style	Description
comic	Classic colorful comic book / webcomic (default)
manga	Japanese manga with speed lines, screen tones
cyberpunk	Dark futuristic with neon colors, glitch effects
retro	1960s pop art, halftone dots, bold primary colors
pixel	16-bit pixel art, limited color palette

Color defaults are applied per style but can be overridden — if your description mentions colors ("black and white", "pastel", "neon"), the default is skipped.

Reference images

The --ref flag enables style consistency across multiple generations. The model matches:

• Line weight and outlining technique
• Coloring style and palette
• Character design and proportions
• Panel framing and composition

This is useful for creating comic series with a consistent look, or matching an existing illustration style.

Backend

Both backends use the same underlying model (Nano Banana Pro / gemini-3-pro-image-preview). If both API keys are configured, the primary backend is tried first. On failure, the other is used as fallback.

	Google Gemini (default)	Kie.ai
API Key	GEMINI_API_KEY	KIE_API_KEY
Pricing	Google AI free tier / pay-as-you-go	18 credits (~$0.09) / image

License

MIT