Creative Image Remix

Problem

When an image already has a clear subject, composition, and visual asset, but the next requirement is to change style, scene, series, material, or cartoonization, manually rebuilding prompts is brittle and loses subject consistency. creative-image-remix turns these creative branches into a reusable workflow, reducing repeated subject and style description overhead.

How It Works

The skill starts with intent routing: if the user has an image and a clear instruction, it goes directly to the matching capability; if the intent is ambiguous, it analyzes the image and proposes options; if no image is provided, it asks for a description or reference. It then builds a prompt from the selected capability module, calls image_gen, and enters an iterate loop for accept, tweak, or redo.
Core capabilities include style transfer (style), reference-based creation (reference), scene replacement (scene), series expansion (series), style fusion (fusion), material conversion (material), era shift (era), composition restructure (compose), lineart conversion (lineart), and IP cartoonization (cartoon). It also handles compound requests, such as style transfer followed by seasonal series, or reference style followed by LEGO material conversion, executing dependent steps in order.

Boundaries

This skill is oriented toward creative image direction, not general retouching or video production. Output quality depends on the underlying image_gen model; complex subjects, precise text, and strict brand guidelines still need human review. By default, it avoids adding glow, drop shadows, or floating shadows unless explicitly requested.