FastResize

**The fastest image resizing library on the planet.** 🚀 Resize 1,000 images in **2 seconds**. Up to **2.9x faster** than libvips, **3.1x faster** than imageflow. Uses **3-4x less RAM** than alternatives. [](https://github.com/tranhuucanh/fast_resize/stargazers) [](https://github.com/tranhuucanh/fast_resize/network/members) [](https://deepwiki.com/tranhuucanh/fast_resize) [Installation](#-installation) • [Quick Start](#-quick-start) • [Documentation](#-documentation) • [Benchmarks](#-performance-benchmarks) --- ### 🛠️ Tech Stack & Stats [](LICENSE) [](https://isocpp.org/) [](https://cmake.org/) [](https://www.gnu.org/software/bash/) 💎 **Ruby:** [](https://rubygems.org/gems/fast_resize) [](https://rubygems.org/gems/fast_resize) 📦 **CLI:** [](https://github.com/tranhuucanh/fast_resize/releases)

---

📑 Table of Contents

• Why Fastest on the Planet?
• Performance Benchmarks
• Key Features
• Installation
• Quick Start
• API Reference
• Documentation
• Architecture
• License
• Contributing
• Contact & Support

---

🏆 Why Fastest on the Planet?

**FastResize has defeated the two most legendary image processing libraries in the world:**

Library	Description	FastResize Advantage
libvips	The gold standard for high-performance image processing. Used by Sharp (Node.js), Shopify, and Wikipedia. Widely recognized as one of the fastest image processing libraries available.	FastResize is 1.7x - 2.9x faster
imageflow	A Rust-based image manipulation engine designed for high throughput and accuracy. Known for powering modern large-scale image resizing pipelines.	FastResize is 1.1x - 3.1x faster

---

🔥 Performance Benchmarks

**Resize 1,000 images to 800px width** *Tested on MacBook Pro M2, macOS Sonoma* | Format | FastResize | libvips (parallel) | imageflow | 🚀 vs libvips | 🚀 vs imageflow | |--------|------------|-------------------|-----------|---------------|-----------------| | **JPG** | **2.10s** ⚡ | 5.24s | 6.60s | **2.5x faster** | **3.1x faster** 🏆 | | **PNG** | **3.43s** ⚡ | 6.18s | 8.41s | **1.8x faster** | **2.5x faster** | | **BMP** | **1.65s** ⚡ | 4.72s | ❌ N/A | **2.9x faster** | - | | **WEBP** | **14.03s** ⚡ | 23.52s | 15.69s | **1.7x faster** | **1.1x faster** | **FastResize wins in ALL formats!** 🎯

📊 Speed Comparison Details

### JPG Performance (1179×1409, 3 channels) | Tool | Run 1 | Run 2 | Avg Time | Throughput | |------|-------|-------|----------|------------| | **FastResize** | 2.02s | 2.17s | **2.10s** | **~477 img/s** ⚡ | | libvips (parallel) | 5.20s | 5.28s | 5.24s | ~191 img/s | | libvips (sequential) | 16.32s | 15.91s | 16.11s | ~62 img/s | | imageflow | 6.62s | 6.58s | 6.60s | ~151 img/s | ### PNG Performance (2104×1160, 4 channels) | Tool | Run 1 | Run 2 | Avg Time | Throughput | |------|-------|-------|----------|------------| | **FastResize** | 3.55s | 3.32s | **3.43s** | **~292 img/s** ⚡ | | libvips (parallel) | 6.18s | 6.18s | 6.18s | ~162 img/s | | libvips (sequential) | 31.95s | 31.46s | 31.71s | ~32 img/s | | imageflow | 8.45s | 8.37s | 8.41s | ~119 img/s | ### BMP Performance (1045×1045, 3 channels) | Tool | Run 1 | Run 2 | Avg Time | Throughput | |------|-------|-------|----------|------------| | **FastResize** | 1.65s | 1.64s | **1.65s** | **~606 img/s** ⚡ | | libvips (parallel) | 4.75s | 4.70s | 4.72s | ~212 img/s | | imageflow | ❌ | ❌ | Not supported | - | ### WEBP Performance (4275×2451, 3 channels) | Tool | Run 1 | Run 2 | Avg Time | Throughput | |------|-------|-------|----------|------------| | **FastResize** | 13.99s | 14.07s | **14.03s** | **~71 img/s** ⚡ | | libvips (parallel) | 23.94s | 23.10s | 23.52s | ~43 img/s | | imageflow | 15.64s | 15.73s | 15.69s | ~64 img/s |

💾 RAM Usage Comparison

FastResize uses **3-4x less RAM** than alternatives! ### PNG (1000 images, 2104×1160) | Library | Time | Throughput | RAM Peak | Compared | |---------|------|------------|----------|----------| | **FastResize** | 3.19s | 313 img/s | **138 MB** ⚡ | - | | libvips | 6.11s | 164 img/s | 439 MB | 3.2x more | | imageflow | 8.39s | 119 img/s | 514 MB | 3.7x more | ### JPG (1000 images, 1179×1409) | Library | Time | Throughput | RAM Peak | Compared | |---------|------|------------|----------|----------| | **FastResize** | 2.03s | 493 img/s | **271 MB** ⚡ | - | | libvips | 5.07s | 197 img/s | 460 MB | 1.7x more | | imageflow | 6.59s | 152 img/s | 541 MB | 2.0x more |

Full Benchmark Details →

---

💪 Key Features

### ⚡ Performance - **Up to 3.1x faster** than libvips & imageflow - **3-4x less RAM** usage - Optimized C++ core with SIMD - Multi-threaded batch processing	### 🖼️ Format Support - **JPEG** (via libjpeg-turbo) - **PNG** (via libpng) - **WebP** (via libwebp) - **BMP** (via stb_image)
### 🎯 Resize Modes - **Exact size** - Fixed width & height - **Fit width** - Auto height - **Fit height** - Auto width - **Scale percent** - Proportional resize	### 🔧 Quality Filters - **Mitchell** - Balanced (default) - **Catmull-Rom** - Sharp edges - **Box** - Fast, lower quality - **Triangle** - Bilinear

---

📦 Installation

### 💎 Ruby ```bash gem install fast_resize ``` ### 🍎 macOS (Homebrew) ```bash brew tap tranhuucanh/fast_resize brew install fast_resize ``` ### 🐧 Linux ```bash # Download latest release VERSION="1.0.0" wget https://github.com/tranhuucanh/fast_resize/releases/download/v$VERSION/fast_resize-$VERSION-linux-x86_64.tar.gz tar -xzf fast_resize-$VERSION-linux-x86_64.tar.gz sudo cp fast_resize /usr/local/bin/ sudo chmod +x /usr/local/bin/fast_resize # Verify installation fast_resize --version ```

---

🎯 Quick Start

CLI

# Resize to width 800 (auto height)
fast_resize input.jpg output.jpg 800

# Resize to exact 800x600
fast_resize input.jpg output.jpg 800 600

# Batch resize all images in directory
fast_resize batch input_dir/ output_dir/ --width 800

# Convert format (JPG → PNG)
fast_resize input.jpg output.png 800

💎 Ruby

require 'fast_resize'

# Simple resize (width 800, auto height)
FastResize.resize('input.jpg', 'output.jpg', width: 800)

# Exact dimensions
FastResize.resize('input.jpg', 'output.jpg', width: 800, height: 600)

# With quality and filter options
FastResize.resize('input.jpg', 'output.jpg',
  width: 800,
  quality: 90,
  filter: :catmull_rom
)

# Batch resize (multi-threaded)
files = Dir['images/*.jpg']
result = FastResize.batch_resize(files, 'output/', width: 800)
puts "Processed: #{result[:success]}/#{result[:total]}"

# Maximum speed mode (uses more RAM)
FastResize.batch_resize(files, 'output/',
  width: 800,
  max_speed: true
)

Full Ruby Guide → • Full CLI Guide →

---

📖 API Reference

Resize Options

Option	Type	Default	Description
width	int	-	Target width in pixels
height	int	-	Target height in pixels
scale	float	-	Scale factor (0.5 = 50%)
quality	int	85	JPEG/WebP quality (1-100)
keep_aspect_ratio	bool	true	Maintain aspect ratio
filter	symbol	:mitchell	Resize filter algorithm

Resize Modes

Mode	When	Example
Fit Width	Only width specified	width: 800 → height auto
Fit Height	Only height specified	height: 600 → width auto
Exact Size	Both specified	width: 800, height: 600
Scale	scale specified	scale: 0.5 → 50% size

Filter Options

Filter	Quality	Speed	Best For
:mitchell	★★★★☆	★★★☆☆	General use (default)
:catmull_rom	★★★★★	★★☆☆☆	Sharp edges, text
:triangle	★★★☆☆	★★★★☆	Smooth gradients
:box	★★☆☆☆	★★★★★	Maximum speed

Batch Options

Option	Type	Default	Description
threads	int	auto	Number of threads (0 = auto)
stop_on_error	bool	false	Stop on first error
max_speed	bool	false	Enable pipeline mode (faster, uses more RAM)

---

📚 Documentation

Complete usage guides:

• CLI Usage Guide - Complete command-line reference
• Ruby Usage Guide - Ruby integration and examples
• Benchmark Details - Full performance analysis
• API Reference - Complete API documentation

---

🏗️ Architecture

FastResize is built on industry-standard, high-performance libraries:

Library	License	Purpose
libjpeg-turbo	BSD-3-Clause	JPEG decode/encode with SIMD
libpng	libpng license	PNG decode/encode
libwebp	BSD-3-Clause	WebP decode/encode
stb_image	Public Domain	BMP and fallback decode
stb_image_resize2	Public Domain	High-quality resize with SIMD

Why So Fast?

• ⚡ SIMD Optimization - SSE2/AVX (x86) and NEON (ARM) acceleration
• 🚀 Zero-copy Pipeline - Minimal memory allocation
• 💪 Multi-threaded - Parallel batch processing
• 🔥 Memory-mapped I/O - Efficient file reading
• 🎯 libjpeg-turbo - 2-6x faster JPEG than standard libjpeg

---

📄 License

FastResize is licensed under the BSD-3-Clause License.

See LICENSE for full details.

---

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

---

🐛 Bug Reports

Found a bug? Please open an issue with:

• Your OS and version
• FastResize version
• Steps to reproduce
• Expected vs actual behavior

---

💖 Support This Project

If FastResize helps you save time and money:

• ⭐ Star this repository
• 🐛 Report bugs and suggest features
• 📖 Improve documentation
• 💬 Share FastResize with others

---

📮 Contact & Support

GitHub: @tranhuucanh • Issues • Discussions

---

🙏 Acknowledgments

Built with: libjpeg-turbo • libpng • libwebp • stb by Sean Barrett

---

**Made with ❤️ by FastResize Project** *If FastResize saves you time, give us a star!* ⭐ [](https://star-history.com/#tranhuucanh/fast_resize&Date) [⬆ Back to top](#fast_resize)