Upscayl brings AI-powered image upscaling to the desktop as a native, open-source application with over 44,000 GitHub stars. Unlike cloud-based upscaling services that require uploading your images to third-party servers, Upscayl processes everything locally using your GPU through the Vulkan graphics API, ensuring broad hardware compatibility across NVIDIA, AMD, and Intel GPUs. The application uses Real-ESRGAN neural network models to intelligently reconstruct missing details in low-resolution images.
The tool offers multiple upscaling models optimized for different types of content — general photography, digital art, anime-style illustrations, and more. Users can also load custom NCNN models for specialized use cases. Batch processing support allows upscaling entire folders of images at once, which is particularly useful for game asset pipelines, web design workflows, or preparing marketing materials. The interface is clean and straightforward, requiring just a few clicks to upscale any image.
Built with Electron and TypeScript, Upscayl is available through official installers for all major desktop platforms, plus community-maintained packages via Homebrew, Flatpak, Snap, and AUR. For developers working with visual assets, Upscayl provides a practical tool for improving image quality in projects where high-resolution source files are unavailable, from upscaling legacy icons and textures to enhancing screenshots and documentation images.