Imagen-Qwen — AI Image Generation with Chinese Text Excellence

中文文档

A Claude Code / OpenClaw skill for AI image generation using Alibaba Cloud Bailian API.

Why This Skill?

Key advantages:

• Best Chinese text — Qwen-Image is one of the best models for rendering Chinese text on images
• Realism + art — Wan series covers everything from quick drafts to professional-grade output
• Ready to use — pip install two packages + one API key to get started

Features

• Qwen-Image 2.0: Latest flagship, native 2K, professional typography rendering
• Qwen-Image legacy: Lighter Chinese/English text rendering models
• Wan Series: Photorealistic images and photography-grade visuals, Wan2.7 supports 4K
• Multiple size presets: 1:1, 16:9, 9:16, 4:3, 3:4, plus 1K/2K/4K
• Cross-platform: Windows, macOS, Linux support
• Multiple API regions: China (default), Singapore, Virginia

Install the Skill

Claude Code (global):

git clone https://github.com/Agents365-ai/imagen-qwen.git ~/.claude/skills/imagen-qwen

Claude Code (project-specific):

git clone https://github.com/Agents365-ai/imagen-qwen.git .claude/skills/imagen-qwen

OpenClaw:

git clone https://github.com/Agents365-ai/imagen-qwen.git skills/imagen-qwen

SkillsMP: Search imagen-qwen on skillsmp.com for one-click install.

Requirements

System Requirements

• Python 3.8+
• pip

Install Dependencies

pip install dashscope requests

API Key

Get your API key from Alibaba Cloud Bailian Console

export DASHSCOPE_API_KEY="your_api_key"

Optional Environment Variables

# Set default model (default: qwen-image-2.0-pro)
export DASHSCOPE_MODEL="wan2.7-image-pro"

# Set API endpoint (default: cn)
export DASHSCOPE_API_BASE="cn"  # or "sg", "us", or full URL

Quick Start

Natural Language (Claude Code)

Just tell Claude what you want:

Generate an image of a cute orange cat
Create a poster with text "Happy New Year" in Chinese
Make a photorealistic 4K mountain sunset photo using wan2.7-image-pro
Generate a 16:9 landscape wallpaper

Command Line

# Basic usage (default model: qwen-image-2.0-pro, native 2K)
python ~/.claude/skills/imagen-qwen/scripts/generate_image.py "A cute cat" output.png

# Photorealistic 4K with Wan2.7
python ~/.claude/skills/imagen-qwen/scripts/generate_image.py --model wan2.7-image-pro --size 4K "Mountain sunset" photo.png

# Custom size
python ~/.claude/skills/imagen-qwen/scripts/generate_image.py --size 16:9 "Wide landscape" landscape.png

# With negative prompt
python ~/.claude/skills/imagen-qwen/scripts/generate_image.py --negative "blurry" "High quality portrait" portrait.png

# List available models
python ~/.claude/skills/imagen-qwen/scripts/generate_image.py --list-models

Models

Size Presets

Qwen-Image 2.0 (native 2K):

• 1:1 → 2048×2048 (default)
• 16:9 → 2688×1536
• 9:16 → 1536×2688
• 4:3 → 2304×1728
• 3:4 → 1728×2304
• 1K → 1024×1024
• 2K → 2048×2048

Qwen-Image legacy:

• 1:1 → 1328×1328
• 16:9 → 1664×928
• 9:16 → 928×1664
• 4:3 → 1472×1104
• 3:4 → 1104×1472

Wan Series (Wan2.7 also accepts 1K/2K/4K):

• 1:1 → 1024×1024
• 1:1-large → 1280×1280
• 16:9 → 1280×720
• 9:16 → 720×1280
• 4:3 → 1200×900
• 3:4 → 900×1200
• 2:1 → 1440×720

API Endpoints

License

MIT License

Support

If this project helps you, consider supporting the author:

WeChat Pay

Alipay

Buy Me a Coffee

Give a Reward

Author

Agents365-ai

• Bilibili: https://space.bilibili.com/441831884
• GitHub: https://github.com/Agents365-ai