Smart Outfit Nine-Grid Generator

The Problem

Outfit previews and virtual try-on often suffer from the same issue: a single rendered image only covers one pose or scene, while manually generating multiple variants creates repetitive work. Smart Outfit Nine-Grid Generator compresses the input to one portrait image and one style prompt, then produces a 3x3 set of related outfit variations for early exploration.

How It Works

The skill treats image generation as an asynchronous workflow:

• Input validation: portrait_image must be a valid Base64 payload, and prompt should describe category, style, scene, color, and fabric when relevant.
• Safety review: input is checked before generation to keep the pipeline compliant.
• Prompt assembly: the skill assembles a nine-grid generation prompt based on the requested outfit, without extra intent amplification.
• Async generation: the task is submitted and returns only a task_id, without blocking the session.
• Polling: the caller queries status by task_id and receives core progress or data after each poll.
• Splitting: after generating one 1:1 nine-grid image, it is divided into nine independent outfit images.

With need_text=true, each sub-image can include outfit-style labels, which is useful for lookbooks or e-commerce previews.

Boundaries

This skill is aimed at outfit effect previews and batch style exploration, not replacing professional fashion photography or high-precision product rendering. Tasks are asynchronous only, so the caller must poll to completion. Without showapi_appkey, the result is a simulated preview and should not be treated as real generated content. Final quality still depends on portrait quality, prompt specificity, and the underlying image model.