libvips

libvips is a demand-driven, threaded image processing library with low memory requirements and high performance. Originally developed at the University of Southampton in the UK for medical image processing, libvips has evolved into one of the fastest and most memory-efficient image processing libraries available, used in production by major services including Netlify, Cloudflare, Shopify, and Instagram. Unlike traditional image processing libraries that load the entire image into memory before processing, libvips uses a demand-driven (lazy evaluation) architecture: it builds a pipeline of image operations as a directed acyclic graph (DAG), and only evaluates the necessary regions of the image when output is requested. This means libvips can process images that are much larger than available RAM, since it never needs to hold the entire image in memory at once. The library also uses multi-threading to parallelize operations across CPU cores automatically. In benchmarks, libvips typically processes images 3 to 5 times faster than ImageMagick while using 5 to 10 times less memory. Key features include: support for JPEG, PNG, WebP, TIFF, GIF, HEIF/HEIC, AVIF, SVG, PDF, FITS, OpenEXR, and raw image formats, over 300 image processing operations (including resize, crop, rotate, flip, blur, sharpen, color space conversion, thresholding, morphological operations, convolution, and statistical operations), ICC color profile support (for accurate color management across devices), image pyramid generation, metadata reading and writing (EXIF, IPTC, XMP), region-of-interest processing, streaming I/O (reading from and writing to streams without loading entire images), and bindings for C, C++, Python (pyvips), JavaScript (via a WebAssembly port called wasm-vips), Ruby (ruby-vips), PHP (php-vips), Go, Rust, and .NET. As of 2026, libvips has over 11,000 stars.