Text to Video

Problem It Addresses

When writing video prompts, a one-sentence Chinese description often lacks camera language, style constraints, and English phrasing. Sending it directly to a text-to-video model can produce inconsistent framing, unstable motion, or shifted visual style. This skill connects the steps of “description, prompt optimization, and video generation” into a short workflow: the user provides natural language, the system rewrites it into an English prompt suitable for Dreamina, and then calls the video generation interface, reducing manual prompt rewrites and API request assembly.

Workflow and Core Capabilities

• Prompt optimization: Uses DeepSeek to convert Chinese descriptions into high-quality English prompts, automatically inferring parameters such as visual style and camera movement so the input better matches video model expectations.
• Video generation: Calls the Dreamina text-to-video API to generate a 1080p video and returns a link, which is useful for short-video drafts, demo material, or storyboard references.
• Integration surface: Triggered through OpenClaw in Feishu / WeChat, so users do not need to manually copy prompts, check request fields, or track asynchronous generation results in chat.

Scope and Caveats

This skill is suited for quickly turning a rough idea into a previewable video, but it is not a precise control system for character consistency, complex multi-shot storytelling, brand guidelines, or licensed assets. Output depends on the availability, quotas, and generation policies of DeepSeek and Dreamina. The skill returns a video link and does not imply that the file has been automatically downloaded locally.