bear2u/my-skills
Overview
This skill removes Gemini logos, AI watermarks, or other small logos from images using OpenCV inpainting. It provides simple functions to remove a specified rectangular region by coordinates or to target common corners (top/bottom, left/right). The goal is quick, local cleanup for images where logos are small and backgrounds are reasonably uniform.
How this skill works
The skill reads the image with OpenCV, builds a binary mask for the logo area, and applies OpenCV inpainting (Telea method) to fill the masked region from surrounding pixels. You can supply exact coordinates, or use corner-based ratios and padding to automatically target typical logo positions. Small masks and appropriate inpainting radius produce the best results.
When to use it
- Remove a Gemini logo or similar small watermark in the image corner.
- Clean up AI-generated image markers when the logo covers a small, uniform background area.
- Prepare images for presentations, mockups, or testing when a quick local edit is acceptable.
- When you can manually estimate coordinates or use corner-based ratios for logo placement.
Best practices
- Prefer small mask areas; inpainting quality drops as masked area grows.
- Inspect image dimensions first and adjust w_ratio/h_ratio or coordinates accordingly.
- Use padding to avoid trimming visible content; iterate with different radius values (e.g., 3–7).
- Save results to a separate output path and keep the original image unchanged.
- Test both INPAINT_TELEA and INPAINT_NS if results are unsatisfactory.
Example use cases
- Remove a Gemini star logo located near bottom-right by specifying coordinates or using bottom_right corner with w_ratio/h_ratio.
- Batch-clean images with consistent logo placement by scripting corner-based removal and fixed ratios.
- Quickly remove a small watermark before composing a presentation slide.
- Repair a stamped AI marker on product photos where the background is simple and uniform.
FAQ
No. Inpainting works best on small regions and relatively uniform backgrounds. Complex or large logos may leave artifacts.
Which inpainting method should I use?
INPAINT_TELEA is a good default. If results are poor, try INPAINT_NS or adjust mask size and radius.
16 skills
This skill analyzes project context from uploaded files and generates enhanced, implementation-ready prompts for development tasks.
This skill performs DuckDuckGo web searches for text, news, and images, returning concise results suitable for programmatic processing.
This skill automatically generates structured, parallel-friendly custom slash commands from high-level prompts, including workflows, dependencies, tests, and
This skill initializes a Flutter project with Clean Architecture, Riverpod, Drift, and modern stack for Todo/Habit/Note/Expense or Custom domains.
This skill orchestrates a triple-AI engineering loop among Claude, Codex, and Cursor to plan, validate, implement, and refine code quality.
This skill generates optimized Midjourney prompts for 600x600 card news backgrounds tailored to topic, mood, and style preferences.
This skill orchestrates a dual-AI engineering loop where Claude Code plans and Codex validates, delivering high-quality, well-documented code.
This skill helps you initialize a Next.js 15 App Router project with Todo/Blog/Dashboard/E-commerce options, modern stack, and automated boilerplate.
This skill generates a 5-7 card Instagram-style card news series with topic input, colors, and optional background images, handling text wrapping.
This skill automatically generates comprehensive workthrough documentation after development tasks, detailing context, changes, and verification to streamline
This skill removes Gemini logos or watermarks from images using inpainting with OpenCV to produce clean results.
This skill helps you create high-converting landing pages with Next.js 14+, TypeScript, Tailwind, and ShadCN UI following 11 essential elements.
This skill generates domain-aware, user-journey oriented design prompts for AI web builders with emotional design insights and actionable guidance.
This skill automatically generates 600x600 Instagram-style card news series with optional background images from user-provided topics, colors, and visuals.
This skill creates distinctive, high-converting landing pages using Next.js 14+ and ShadCN UI, delivering memorable design and production-ready code.
This skill generates code explanation cards with title and explanation sections, including syntax highlighted code samples and Korean explanations for
