Vidu Q2 Image Generation

What It Solves

When you need to turn a prompt or reference image into an image without wiring model calls, auth, async tasks, and file uploads into your own scripts, this skill gives you a lightweight entry point around the dlazy viduq2-t2i command. It targets Vidu Q2 image generation, covering text-to-image and image-to-image requests, and fits naturally into an agent workflow as an image-generation node.

How It Works

The skill is a thin client: your prompt, parameters, and local file paths are sent through the dLazy CLI to api.dlazy.com. If you pass local image, video, or audio paths, the CLI uploads them to files.dlazy.com so the hosted model can read them. After generation, the API returns output URLs hosted on files.dlazy.com.

Key steps include:
- Configure DLAZY_API_KEY or store the key via the dLazy CLI
- Run dlazy viduq2-t2i with a prompt and other parameters
- Use --no-wait for async tasks, which returns a generateId, then poll with dlazy status --wait
- Handle errors such as unauthorized, insufficient_balance, and backend safety-policy failures

Boundaries

This is a SaaS generation client, not a local inference engine. Beyond what the dLazy CLI already handles for auth, uploads, and requests, the skill does not perform additional network or filesystem access. Treat API keys as secrets and keep in mind that inputs and outputs are subject to dLazy service terms.