lxfater/nano-image-generator-skill

Overview

This skill generates high-quality images using Nano Banana Pro (Gemini 3 Pro Preview) for app icons, logos, UI graphics, banners, illustrations, and diagrams. It supports reference images to transfer visual style or preserve character consistency across outputs. Use simple CLI commands to produce single images or batch series with consistent aesthetics. Outputs can be configured for aspect ratio and resolution.

How this skill works

You provide a text prompt and optional reference images; the tool renders an image with the requested composition, style, and resolution. Reference images (up to 14) enable style transfer, character consistency, or multi-image fusion by telling the model to match colors, textures, or subjects. Options let you set aspect ratio and size (1K, 2K, 4K) and produce serial batches by reusing a generated image as a reference. Generated files are saved to the specified output path.

When to use it

• Creating app icons or logo concepts with a specific visual style
• Designing marketing banners, hero images, or social media posts
• Producing UI graphics, illustrations, or diagrams for product screens
• Generating series of images where consistent style or character continuity is required
• Quickly prototyping visual assets for pitches, mockups, or A/B tests

Best practices

• Write specific prompts including style, purpose, colors, and composition
• Use reference images and explicit phrases like 'same style as reference' to preserve aesthetic
• Choose aspect ratio and resolution that match the final use (icons 1:1, banners 16:9)
• Generate a master image first and reuse it as a reference for consistent series
• Avoid requesting dense readable text inside images—overlay text in your editor

Example use cases

• Generate a minimalist flat-design app icon at 1:1 for mobile stores
• Create a website hero banner in 16:9 with abstract shapes and brand colors
• Transform a photo into a watercolor-style illustration using a style reference
• Produce multiple knowledge cards with a consistent cover style using serial generation
• Recreate a character from a reference image in new scenes while keeping appearance