tensorslab-image

What This Skill Does

The tensorslab-image skill acts as a powerful interface between your OpenClaw agent and TensorsLab's advanced AI image generation suite. This skill allows users to generate high-fidelity visuals from text descriptions, transform existing images through image-to-image workflows, and perform complex image manipulations such as object erasure, face replacement, and watermark removal. The skill automatically handles prompt engineering by enriching user requests with specific details regarding lighting, composition, and style, ensuring consistent high-quality outputs. All generated files are automatically managed and saved to a local output directory for immediate access.

Installation

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

clawhub install openclaw/skills/skills/bob5-tensorslab/tl-image

Before running the skill, ensure you have set your API key by exporting it as an environment variable: export TENSORSLAB_API_KEY="your-key-here". You can verify this configuration at any time by checking the agent's internal configuration logs or running a standard shell check for the presence of the environment variable.

Use Cases

This skill is designed for diverse workflows including:

• Creative Content Production: Generating artistic or photorealistic imagery from natural language prompts.
• Asset Modification: Using advanced editing tools to alter specific elements within an image, such as changing sky conditions, hair color, or adding accessories to subjects.
• Automated Design Tasks: Performing batch-style tasks like removing watermarks or object erasure without manual image editing software.
• Transformation Projects: Converting sketches or basic wireframes into fully rendered 3D-styled models using image-to-image capabilities.

Example Prompts

1. "画一个在月球上吃热狗的宇航员"
2. "把 cat.png 的背景换成太空，保持猫咪的姿态不变"
3. "把这张照片里人物的头发颜色染成粉色，并加上墨镜"

Tips & Limitations

• Model Selection: The skill defaults to seedreamv4 for most tasks. For high-end creative work, experiment with seedreamv45 for enhanced detail and zimage for specific artistic styles.
• Prompt Quality: While the agent automatically enhances your prompts, providing specific details about the desired mood, composition, or time of day in your initial request often yields better results.
• Resource Management: Ensure you have enough disk space for the ./tensorslab_output/ folder, as high-resolution images can grow in size rapidly. If the agent reports a failure, confirm your API key is correctly exported and that you have sufficient credits on the TensorsLab console.