vgl

What This Skill Does

The vgl skill enables precise, deterministic control over AI image generation by utilizing Bria's Visual Generation Language (VGL). Instead of relying on the unpredictable nature of natural language prompts, this skill forces a structured JSON output that defines every pixel-critical aspect of an image. It acts as a bridge between high-level user intent and the low-level mechanical requirements of Bria FIBO models, ensuring that composition, object placement, lighting, and camera physics are explicitly defined.

Installation

You can install the skill directly from the OpenClaw repository using the following command: clawhub install openclaw/skills/skills/galbria/vgl

Use Cases

This skill is designed for scenarios where visual consistency and reproducibility are mandatory.

1. Brand Identity: Ensure your subject is always positioned in the bottom-left corner with specific studio lighting for social media assets.
2. Product Photography: Define exact object textures and relative scaling to maintain a uniform look across a product catalog.
3. Complex Composition: Create scenes with multiple, specific subjects that natural language models often fail to place correctly, such as specific interactions or directional orientations.
4. Batch Generation: Programmatically generate high-fidelity variations of a base image by modifying specific JSON attributes while keeping the rest of the scene locked.

Example Prompts

1. "Generate a VGL JSON for a professional portrait of a woman wearing a blue silk blouse, studio lighting, front-lit, large within frame."
2. "I need a deterministic image of a futuristic desk setup. Place a mechanical keyboard in the center and a laptop to the left. Set the lighting to cool blue neon with sharp shadows."
3. "Convert this description to VGL format: An old, weathered stone bridge over a misty river at golden hour with soft, long shadows."

Tips & Limitations

• Precision is Key: The more specific you are in your natural language input, the more robust the generated VGL JSON will be. Use modifiers like 'large within frame' to control framing.
• Object Constraints: You are limited to a maximum of 5 objects per generation. If your scene requires more, consider generating elements in layers.
• Integration: Always pair this with the bria-ai skill to execute the structured JSON prompts. VGL is the blueprint, but Bria is the engine that renders the final file.
• Validation: Always verify the JSON keys before passing them to an execution agent; invalid schema will cause generation failures.