Upscayl - AI Image Tools Tool

Overview

Upscayl is a free, open-source desktop application for AI-powered image upscaling and enhancement. Built with an Electron/TypeScript front end and an NCNN/Vulkan backend, Upscayl runs locally on Linux, macOS and Windows and uses Real-ESRGAN–derived models to reconstruct detail when enlarging images. The app exposes several pre-trained models (for example “High Fidelity”, “Ultrasharp”, and community models), batch processing, and a two-pass “Double Upscayl” workflow for higher effective magnification. According to the project site and repository, Upscayl intentionally prioritizes a Linux-first development approach while providing distribution formats for the major desktop platforms. ([github.com](https://github.com/upscayl/upscayl)) The project is actively maintained and packaged for end users (AppImage/Flatpak/DEB/RPM for Linux, DMG/Homebrew for macOS, and an NSIS/MSI-style installer for Windows). Upscayl requires a Vulkan‑compatible GPU (NCNN/Vulkan does the heavy lifting), so users should verify Vulkan driver support on their hardware before relying on CPU-only operation. Recent releases (for example v2.15) added a new “High Fidelity” model, multi-language UI strings, a Lens Viewer, and clipboard image upscaling, reflecting continued feature development and maintenance. ([github.com](https://github.com/upscayl/upscayl/releases?utm_source=openai))

GitHub Statistics

• Stars: 42,232
• Forks: 2,014
• Contributors: 46
• License: AGPL-3.0
• Primary Language: TypeScript
• Last Updated: 2025-10-28T10:39:30Z
• Latest Release: v2.15.0

The GitHub repository is large and active: the project has tens of thousands of stars and several thousand forks, is licensed under AGPL‑3.0, and shows regular releases and changelogs on its Releases page. The codebase includes the Electron-based desktop UI and a separate upscayl-ncnn backend (Real-ESRGAN / ncnn Vulkan fork) for inference. Issue and PR activity shows an engaged user base and a small maintainers team; the project accepts community model contributions via a dedicated custom-models repo and maintains a compatibility wiki documenting Vulkan/GPU behavior. Overall community health is strong (large star count and many contributors) but issues around GPU compatibility and occasional crashes appear in support threads. ([github.com](https://github.com/upscayl/upscayl))

Installation

Install via brew:

macOS (Homebrew): brew install --cask upscayl

macOS (manual): Download upscayl-<version>-mac.dmg from Releases, open the .dmg and drag to /Applications.

Linux (portable AppImage): chmod +x upscayl-<version>-linux.AppImage && ./upscayl-<version>-linux.AppImage

Linux (Flatpak/AUR/Snap): install from your distro store or via Flatpak/AUR packages (see Releases/wiki).

Windows (installer): Download upscayl-<version>-win.exe from Releases and run the installer.

From source (dev): git clone https://github.com/upscayl/upscayl && cd upscayl && npm install && npm run start

Package from source: npm run dist (creates platform packages via electron-builder).

Key Features

• Local, offline upscaling—no cloud upload of images required.
• NCNN + Vulkan inference backend (Real-ESRGAN family models) for GPU-accelerated upscaling.
• Multiple built-in models (High Fidelity, Ultrasharp, Remacri, etc.) for different content types.
• Batch Upscayl: queue and process many images in one operation.
• Double Upscayl: two-pass upscaling workflow for higher effective magnification.
• Custom model support via the custom-models repository.
• Clipboard paste upscaling and a Lens Viewer for side-by-side comparisons.
• Export options include original formats and WEBP; supports 2×/4× passes and sequential upscales.

Community

Upscayl has a large, active open-source community (many tens of thousands of GitHub stars, forks, and dozens of contributors) and an active Releases/Issues workflow; the project maintains a compatibility wiki and a separate upscayl-ncnn backend repo for model inference. Users generally praise the app as a free, effective local upscaler—especially for digital art and older photos—but community threads also call out GPU/Vulkan driver headaches, occasional instability, and mixed results on heavily blurred inputs. The desktop app is free under AGPL‑3.0, while some platform storefront builds (e.g., Mac App Store) have paid storefront variants and extras. ([github.com](https://github.com/upscayl/upscayl))

Key Information

• Category: Image Tools
• Type: AI Image Tools Tool