Professional-grade AI image upscaler powered by Vulkan compute shaders
V-Upscale is a cross-platform desktop application that uses advanced interpolation algorithms and GPU acceleration to upscale images with exceptional quality. Built with Rust, Tauri, and React for maximum performance and modern UI. I was already using this in a bigger application and decided to make it standalone so brought over some of the code here. hope you enjoy!
🎯 Just Download & Run!
Unlike other GPU-accelerated image tools, V-Upscale requires zero setup - no Vulkan SDK, no driver updates, no configuration. Everything is bundled internally for a seamless user experience.
-
🎨 Professional Quality Upscaling
- Mitchell-Netravali bicubic interpolation for 2x/4x scaling
- Lanczos-2 filter for maximum sharpness at 4x+ scaling
- Eliminates pixelation and blurring common in traditional upscalers
-
⚡ GPU Acceleration
- Vulkan compute shaders for lightning-fast processing
- MoltenVK integration for native macOS support
- Utilizes dedicated GPU memory for large images
-
🔧 Advanced Post-Processing
- Smart sharpening with edge detection
- Adaptive contrast and saturation enhancement
- Optional noise reduction for vintage images
-
🎛️ Multiple Upscaling Methods
- Enhanced: Professional quality with post-processing (recommended)
- Nearest Neighbor: Pixel-perfect for pixel art and testing
- Legacy: Basic upscaling for comparison
-
🖥️ Zero-Dependency Installation
- No Vulkan SDK required - Everything bundled internally
- No driver installations - Works out of the box
- Self-contained binaries - Just download and run
- Native apps for macOS, Windows, and Linux
| Original (640×480) | V-Upscale 4x Enhanced (2560×1920) |
|---|---|
| Upload your image | Professional restoration quality |
Supported formats: PNG, JPEG, WEBP, BMP
Output format: PNG (lossless)
V-Upscale is designed to work immediately without any setup:
- ✅ No Vulkan SDK installation - All libraries bundled internally
- ✅ No driver updates required - Works with existing GPU drivers
- ✅ No configuration needed - Optimized settings out of the box
- ✅ Portable installation - Can run from any location
- ✅ Offline operation - No internet connection required
Perfect for users who want professional image upscaling without the technical hassle.
🌐 Visit our website: https://augani.github.io/v-upscale/
- Download the latest release from Releases
- Install the appropriate package for your platform:
- macOS: Open the
.dmgand drag to Applications - Windows: Run the
.msiinstaller - Linux: Install the
.debpackage or run the.AppImage
- macOS: Open the
- Launch V-Upscale - No additional setup required!
- Optional: Test GPU acceleration with the "Test MoltenVK" button
- Select Image: Click the upload area or drag & drop
- Choose Factor: Select 2x, 4x, or 8x upscaling
- Upscale: Click "Upscale Image" and wait for processing
- Download: Save the enhanced result to your desired location
Note for Users: You don't need any of this! Just download the pre-built app. This section is only for developers who want to build from source.
- Rust (latest stable) - Install Rust
- Node.js (18+) - Install Node.js
- Vulkan SDK - Install Vulkan SDK (bundled in release builds)
# Install Vulkan SDK via Homebrew
brew install vulkan-headers vulkan-loader vulkan-tools
# Install MoltenVK (included in Vulkan SDK)
# Verify installation
vulkaninfo# Clone the repository
git clone https://github.com/augani/v-upscale.git
cd v-upscale
# Install dependencies
npm install
# Setup Rust/Tauri dependencies
cd src-tauri
cargo build
cd ..# Start development server
npm run tauri:dev
# Build frontend only
npm run build
# Build release version
npm run tauri:build
# Test with Vulkan debug mode
npm run dev:vulkan- Frontend: React + TypeScript + Vite
- Backend: Rust + Tauri
- GPU Compute: Vulkan GLSL compute shaders
- Image Processing:
image+imageproccrates - Graphics API: Ash (Vulkan bindings for Rust)
graph TD
A[Input Image] --> B[Load to GPU Memory]
B --> C{Upscale Factor}
C -->|2x-4x| D[Mitchell-Netravali Bicubic]
C -->|4x+| E[Lanczos-2 Filter]
D --> F[Vulkan Compute Shader]
E --> F
F --> G[GPU Processing]
G --> H[Download from GPU]
H --> I{Post-Processing}
I -->|Enabled| J[Noise Reduction]
J --> K[Contrast Enhancement]
K --> L[Smart Sharpening]
I -->|Disabled| M[Output Image]
L --> M
The core upscaling logic runs in GLSL compute shaders:
- Workgroup Size: 16×16 threads for optimal GPU utilization
- Memory Access: Efficient texture sampling with bounds checking
- Interpolation: Real-time bicubic/Lanczos computation
- Precision: 32-bit floating-point for quality preservation
// Enhanced upscaling with post-processing options
upscale_image_enhanced(
path: String, // Input image path
factor: u32, // 2, 4, or 8
apply_sharpening: Option<bool>, // Default: true
apply_contrast_enhancement: Option<bool>, // Default: true
apply_noise_reduction: Option<bool> // Default: false
) -> Result<String, String>
// Pixel-perfect nearest neighbor upscaling
upscale_image_nearest_neighbor(
path: String,
factor: u32
) -> Result<String, String>
// Test MoltenVK setup and Vulkan initialization
test_moltenvk_setup() -> Result<String, String>import { invoke } from "@tauri-apps/api/core";
// Enhanced upscaling
const result = await invoke<string>("upscale_image_enhanced", {
path: "/path/to/image.jpg",
factor: 4,
apply_sharpening: true,
apply_contrast_enhancement: true,
apply_noise_reduction: false
});
// Test MoltenVK
const status = await invoke<string>("test_moltenvk_setup");npm run tauri:dev# Build optimized release
npm run tauri:build
# Output locations:
# macOS: src-tauri/target/release/bundle/macos/v-upscale.app
# DMG: src-tauri/target/release/bundle/dmg/v-upscale_*.dmg# Modify src-tauri/shaders/upscale.comp
# Recompile shader
cd src-tauri/shaders
glslc upscale.comp -o upscale.spv| Image Size | Method | Processing Time | Quality Score |
|---|---|---|---|
| 640×480 → 2560×1920 | Enhanced 4x | ~2-5 seconds | ⭐⭐⭐⭐⭐ |
| 1024×768 → 4096×3072 | Enhanced 4x | ~8-15 seconds | ⭐⭐⭐⭐⭐ |
| 320×240 → 2560×1920 | Enhanced 8x | ~3-8 seconds | ⭐⭐⭐⭐ |
Benchmarks on M1 MacBook Pro with 16GB unified memory
Minimum Requirements (everything bundled - no installations needed):
- macOS: 10.15+ with any Metal-capable GPU
- Windows: Windows 10+ with any DirectX 11+ GPU
- Linux: Any recent distribution with basic GPU drivers
Recommended for best performance:
- RAM: 8GB+ (for processing large images)
- Storage: 100MB free space for installation
- GPU: Dedicated graphics card (integrated GPUs work but slower)
No additional software required - V-Upscale includes everything needed internally!
We welcome contributions! Please see our Contributing Guidelines for details.
- Fork the repository
- Create a feature branch (
git checkout -b feature/amazing-feature) - Commit your changes (
git commit -m 'Add amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request
- 🖼️ Additional image formats (TIFF, RAW, etc.)
- 🎨 New interpolation algorithms (ESRGAN, Real-ESRGAN)
- 🐧 Linux support and packaging
- 🪟 Windows DirectX backend
- 📱 Batch processing capabilities
- 🔧 Performance optimizations
"Test MoltenVK" button shows failure:
- Solution: This is usually fine! The app will fall back to software rendering
- On older Macs: GPU acceleration may not be available, but upscaling still works
- Performance: Software rendering is slower but produces identical quality
App won't launch:
- macOS: Try right-clicking and selecting "Open" to bypass Gatekeeper
- Windows: Windows Defender may quarantine the file - add an exception
- Linux: Ensure you have basic graphics drivers installed (
sudo apt install mesa-utils)
Slow processing:
- Close other GPU-intensive applications (games, video editing)
- Use smaller upscale factors for very large images
- Ensure you have sufficient RAM for the output image size
"Out of memory" errors:
- Try a smaller upscale factor (2x instead of 8x)
- Close other applications to free up RAM
- For huge images, consider cropping or splitting the image
Build errors:
# Clean and rebuild
cargo clean
npm run tauri:buildThis project is licensed under the MIT License - see the LICENSE file for details.
- Vulkan SDK - Cross-platform graphics API
- MoltenVK - Vulkan implementation for macOS
- Tauri - Rust-based app framework
- Ash - Vulkan bindings for Rust
- React - Frontend framework
- GitHub Issues - Bug reports and feature requests
- Discussions - Questions and community chat
Made with ❤️ and Vulkan compute shaders