screenshot-compression

Screenshot Compression reduces the file size of PNG and JPEG screenshots in place while preserving the original file format, so the compressed images keep working everywhere. It wraps two established command-line tools, pngquant for PNG and jpegoptim for JPEG, both of which are particularly effective on screenshot content such as UI elements, text, and flat color regions. A Python script (compress_screenshots.py) accepts one or more files or directories, auto-detects each file's format by extension, compresses it, and prints a per-file and total compression summary.

The format-preserving approach is deliberate. WebP compresses better but has gaps in compatibility: it is not supported by the WeChat editor, Word, or PowerPoint, and support varies across forums and blog backends. Keeping PNG and JPEG guarantees the results embed cleanly in GitHub READMEs, documentation, blog posts, and office documents. Typical use cases are shrinking images before committing them to a repository or batch-compressing a directory of screenshots.

Behavior is controlled with a few options: a required positional list of paths, -r/--recursive to descend into directories, --png-quality (default 80-95, a pngquant min-max range) and --jpeg-quality (default 85). Under the hood, pngquant quantizes 24-bit true color down to an 8-bit 256-color palette with Floyd-Steinberg dithering, typically cutting 60 to 80 percent, while jpegoptim re-encodes at the chosen quality, strips metadata via --strip-all, and optimizes Huffman tables for a 20 to 50 percent reduction. A quality tuning guide suggests higher settings for code screenshots that need sharp text and more aggressive settings for thumbnails. Two important caveats apply: files are overwritten in place, so back up originals first if needed, and pngquant and jpegoptim must already be installed. The script checks for them and prints platform-specific install instructions (brew, apt, or yum) if either is missing, but never installs them itself. Only .png, .jpg, and .jpeg files are processed; other formats are silently skipped.