qwen-image

What This Skill Does

The qwen-image skill is a sophisticated image generation interface for OpenClaw that leverages Alibaba Cloud's state-of-the-art Bailian models. It acts as an intelligent router between two specialized model families: Z-Image (optimized for realistic portraits and film-style photography) and Qwen-Image (a powerful general-purpose engine designed for complex artistic compositions, text-heavy designs, and high-quality illustrations). By using this skill, users benefit from automatic scene detection, which selects the best underlying engine based on the prompt's intent, ensuring that skin textures appear natural for photos while creative text renders clearly in graphical designs.

Installation

To integrate this skill into your OpenClaw environment, execute the following command in your terminal:

clawhub install openclaw/skills/skills/irron/qwenz-image-gen

After installation, ensure you have configured your Alibaba Cloud Bailian API credentials. You can set the DASHSCOPE_API_KEY either as a global environment variable or within your project's TOOLS.md file. Please ensure your API key matches the geographic requirements of your deployment region.

Use Cases

• Professional Portraits: Generate high-fidelity human images with authentic film grain and cinematic lighting using the z-image-turbo engine.
• Graphic Design: Produce posters or banners that require precise Chinese character rendering, which the qwen-image-max model handles with industry-leading accuracy.
• Artistic Concept Art: Create detailed illustrations, manga, or oil-painting style images by utilizing the general-purpose qwen-image series.
• Rapid Prototyping: Save time by allowing the agent to automatically infer the model choice, reducing the need for manual parameter tuning.

Example Prompts

1. "Generate a realistic portrait of a young woman standing in a snow-covered field, focus on film grain and Kodak Portra 400 aesthetics."
2. "Create an anime-style illustration of a sci-fi city landscape with the Chinese characters '未来之城' written in bold neon typography at the center."
3. "Design a professional business card background with a minimalist marble texture and elegant gold accents, using a 1:1 aspect ratio."

Tips & Limitations

• Optimization: For the best results, enable --prompt-extend to allow the AI to flesh out your creative vision.
• Character Constraints: Remember that while the models are powerful, your positive prompt should stay under 800 characters, and negative prompts should be kept under 500 characters.
• Model Selection: If you have a specific artistic requirement that the automatic router fails to catch, you can override it using the --model flag (options: z for portraits, qwen for general).
• Regional Note: Keep in mind that API keys for the Beijing and Singapore regions are isolated; ensure your configuration aligns with the region where your account is provisioned.