VIPSKit is a pure Swift wrapper for libvips, the fast image processing library, for Apple platforms. The heavy lifting (compiling libvips + 17 dependencies) is handled by the separate vips-cocoa project, which produces a static vips.xcframework. VIPSKit statically links that xcframework to provide a clean, type-safe Swift API. VIPSKit itself is then vended as a dynamic framework in order to still fulfill the obligations of the LGPL-2.1 license.
libvips is known for being exceptionally fast and memory-efficient, using a streaming architecture that processes images incrementally rather than loading entire images into memory. This makes it ideal for processing large images or batch thumbnail generation on mobile devices.
- Pure Swift API with full async/await support.
- Supports iOS 15+, macOS 12+, and visionOS 1.0+.
- Image format support for JPEG, PNG, WebP, JPEG-XL, TIFF, HEIF, AVIF, and GIF.
- Memory-efficient shrink-on-load thumbnailing for JPEG, WebP, and HEIF images.
- Smart content-aware cropping using attention detection.
- Image compositing with 25 blend modes for watermarks and overlays.
- Color adjustments (brightness, contrast, saturation, gamma).
- Gaussian blur, sharpening, and edge detection (Sobel, Canny).
- Drawing primitives (rectangles, lines, circles, flood fill).
- Image analysis (trim detection, statistics, average/background color).
- Full EXIF, XMP, and ICC metadata access.
- Efficient tiling and region extraction for very large images.
- Direct CGImage export for zero-copy display.
VIPSKit features a simple, expressive API that handles all the complexity of libvips internally.
import VIPSKit
// Initialize once at app start
try VIPSImage.initialize()
// Load and create a thumbnail efficiently (shrink-on-load)
let thumbnail = try VIPSImage.thumbnail(fromFile: path, width: 200, height: 200)
// Or load for full processing
let image = try VIPSImage(contentsOfFile: path)
// Smart crop to find interesting regions
let cropped = try image.smartCrop(toWidth: 400, height: 400, interesting: .attention)
// Color adjustments
let adjusted = try image.adjust(brightness: 0.1, contrast: 1.2, saturation: 1.1)
// Add a watermark
let watermarked = try image.composite(withOverlay: watermark, mode: .over, x: 10, y: 10)
// Export to data or file
let jpegData = try image.data(format: .jpeg, quality: 85)
try image.write(toFile: "/path/to/output.jpg")
// Create CGImage directly for display (most efficient)
let cgImage = try image.cgImage
let uiImage = UIImage(cgImage: cgImage)All I/O and CPU-heavy operations also have async variants:
// Async loading
let image = try await VIPSImage.loaded(fromFile: path)
let thumb = try await VIPSImage.thumbnail(fromFile: path, width: 200, height: 200)
// Async processing
let resized = try await image.resizedToFit(width: 800, height: 600)
let blurred = try await image.blurred(sigma: 2.0)
// Async export
let data = try await image.encoded(format: .webP, quality: 80)VIPSKit supports the following platforms:
| Platform | Architectures | Min Version |
|---|---|---|
| iOS | arm64 | 15.0 |
| iOS Simulator | arm64, x86_64 | 15.0 |
| Mac Catalyst | arm64, x86_64 | 15.0 |
| macOS | arm64, x86_64 | 12.0 |
| visionOS | arm64 | 1.0 |
| visionOS Simulator | arm64 | 1.0 |
Add to your Package.swift:
dependencies: [
.package(url: "https://github.com/TimOliver/VIPSKit.git", from: "1.0.0"),
]Or in Xcode: File > Add Package Dependencies, enter the repository URL.
VIPSKit automatically pulls in the pre-built vips.xcframework from vips-cocoa via SPM binary targets.
- Build
VIPSKit.xcframework(see Building from Source below). - Drag it into your Xcode project.
- Add to "Frameworks, Libraries, and Embedded Content".
- Set "Embed" to "Embed & Sign".
- Xcode 16+ (Swift 6.0+)
- Ruby with
xcodeprojgem:gem install xcodeproj - Static
vips.xcframeworkinFrameworks/(from vips-cocoa)
# Copy vips.xcframework from vips-cocoa
cp -R ~/Developer/vips-cocoa/build/xcframeworks/ios/static/vips.xcframework Frameworks/
# Generate Xcode project
ruby Scripts/configure-project.rb
# Open and run tests (⌘U)
open VIPSKit.xcodeprojswift build # Build
swift test # Run testsBuild the VIPSKit.xcframework:
./build.shThis archives for iOS, iOS Simulator, and Mac Catalyst, then produces VIPSKit.xcframework in the project root.
./build.sh --clean # Clean build artifacts first
./build.sh --fast # Build for current platform only (skip archiving)| Format | Read | Write | Notes |
|---|---|---|---|
| JPEG | Yes | Yes | libjpeg-turbo with SIMD |
| PNG | Yes | Yes | libpng |
| WebP | Yes | Yes | Lossy and lossless |
| JPEG-XL | Yes | Yes | libjxl |
| TIFF | Yes | Yes | Built-in |
| HEIF | Yes | No | Decode via libheif |
| AVIF | Yes | No | Decode via dav1d + libheif |
| GIF | Yes | No | Decode only, built-in |
Traditional image processing libraries load entire images into memory before processing. For a 20MP photo, this can mean 80MB+ of RAM just for the pixel buffer. Multiply that by a few concurrent operations, and mobile devices quickly run into memory pressure.
libvips uses a different approach: it streams pixels through a pipeline, processing only the portions needed at any time. Combined with "shrink-on-load" for formats like JPEG (which can decode directly at reduced resolution), this results in dramatically lower memory usage.
VIPSKit was created by Tim Oliver. libvips is developed by John Cupitt and contributors.
Both libvips and VIPSKit are licensed under the LGPL-2.1.
VIPSKit statically links libvips into the framework binary, which is then vended as a dynamic framework in order to continue fulfilling the LGPL obligations.
See LICENSE for details.
This project was built extensively with the help of Claude Code (Opus 4.6). Given the complexity and breadth of features enabled by libvips, manually implementing every wrapper feature would have been a tremendously time-consuming undertaking otherwise.
All code and build output has been reviewed and tested, but as with any project of this complexity, AI-assisted or not, bugs may exist. If you encounter incorrect behaviour, please open an issue.