AI Cinematic Short Video Director

Problem it solves

Commercial short-video generation often breaks down at the prompt stage: a one-line product brief can produce vague visuals, while conflicting channel, style, and selling points can steer output away from the conversion goal. This skill turns product, channel, and style inputs into a more controlled prompt-engineering workflow.

How it works

It starts with trigger validation and input parsing for Subject/Product, Channel/Intent, and Style/Mood. Missing fields are completed from business logic and marked as [补全推断] instead of asking follow-up questions. It then selects a template such as premium brand outbound, visual seeding, story-driven selling, or B2B industrial tech, and uses event-driven shot logic rather than naive time slices. The final output can include Chinese and English prompts plus relevant parameters, aligned with the wording habits of Sora, Runway Gen-3, Kling, and similar video models.

Boundaries

It generates structured prompts, shot scripts, and copy; it does not render video. It is not intended for pure copywriting, static image generation, full film scripts, or video editing. For real brand marks, portraits, music, or other licensed assets, verify authorization separately, and follow each platform's model limits and documentation.