Nano Banana–like Local Image Reasoning Skill

To maintain character continuity across multiple image generation panels, this Skill enforces a planner-driven approach that parses prompts into a structured, multi-stage plan before executing ComfyUI detector and inpaint loops for panel-level consistency.

A planner-driven local image workflow for multi-panel comics creates an explicit structured plan defining portraits, scenes, and overlays before any generation begins, ensuring character and prop consistency throughout the ComfyUI execution sequence.

Yes, you need local image pipeline components and related models present in your environment, because this Skill orchestrates ComfyUI-based execution relying on detector and inpaint loops for accurate local image generation planning.

To plan a multi-panel story sequence from a single text prompt, the Skill parses your input into a structured panel sequence with defined portraits, scenes, and overlays to preserve character continuity across frames.

You should use detector and inpaint loops for local image generation when you need to improve output fidelity and guarantee panel-level continuity in multi-panel stories orchestrated through the ComfyUI execution engine.