NoirDeleter

Automatically delete grayscale or mostly black-and-white images from a folder.

What it does

Scans a folder recursively and deletes images that exceed a grayscale threshold (0-100%). Uses fast file signatures and SQLite caching to avoid re-processing unchanged files.

Supported formats: JPG, JPEG, PNG, BMP, WebP, GIF, TIF, TIFF

Requirements

pip install Pillow numpy

Usage

Edit the configuration at the bottom of NoirDeleter.py:

process_folder(
    folder_path=r"F:\Porn\ToSort",  # Directory to scan (includes subdirectories)
    threshold=99,                    # Grayscale % to delete (0-100). Recommended: 99
    workers=12                       # Parallel threads (max 12 on HDD)
)

Then run:

python NoirDeleter.py

How it works

1. Calculates a grayscale score (0-100) for each image by comparing RGB channel differences
2. Deletes images meeting or exceeding the threshold
3. Caches results in NoirDeleter.cache.sqlite3 to skip unchanged files on subsequent runs
4. Processes images in parallel using worker threads

Performance

• Cached files are skipped entirely (fastest)
• Speed: 30-150 images/sec depending on image size, about 500/s for cached files (guesstimated on:5600X, sata SSDs, 3600MTs ram)
• Periodic saves prevent excessive memory usage on large folders

Notes

• The script permanently deletes matched files—use with caution
• Threshold of 99 is recommended to catch only nearly-grayscale images
• Results are cached; delete NoirDeleter.cache.sqlite3 to re-evaluate all images