Image Deduplicator

😫 The Pain Point

After multiple backups, your hard drive is full of backup1, copy_of_backup. You have 5,000 photos, but 2,000 are duplicates. Scanning by “Filename” fails because one is named img001.jpg and the other backup_img001.jpg.

🚀 Agentic Solution

Visual Fingerprinting (Hashing): It looks at the content of the image. Even if you resize or rename it, the “fingerprint” remains similar.

Key Features:

• Visual Compare: Shows you the two images side-by-side to verify before deleting.
• Smart Select: Suggests keeping the higher-resolution version.

⚔️ Phase 1: Commander (Quick Fix)

For a quick scan.

Prompt:

“Scan this directory for duplicate images using imagehash (phash algorithm). Print out groups of duplicate files.”

Result: A list of redundant files.

🏗️ Phase 2: Architect (Permanent Tool)

For Clean-up Freaks.

Engineering Prompt:

**Role:** Python AI Developer
**Task:** Create a "Visual Image Deduplicator".
**Requirements:**
1.  **Tech Stack:** `imagehash`, `pillow`, `tkinter`.
2.  **GUI:**
    *   Select Scan Folder.
    *   "Scan" button.
    *   **Review Mode (Crucial):** Display duplicates side-by-side. Checkbox to delete "Image A" or "Image B".
3.  **Logic:**
    *   Compute `phash` for every image.
    *   Store in dict `{hash: [file_list]}`.
    *   If len(list) > 1 -> Duplicate.
4.  **Deliverables:** `dedup_img.py`, `run.bat` (Windows), `run.sh` (Mac).

🧠 Prompt Decoding

• Perceptual Hashing: Unlike cryptographic hashing (MD5) where 1 pixel change alters the whole hash, Perceptual Hashing produces similar hashes for similar looking images. This is the secret sauce.

🛠️ Instructions

1. Copy Prompt -> Paste -> Run.
2. Scan -> Review Matches -> Clean.