Final2x is a cross-platform image super-resolution tool, licensed under BSD-3-Clause, that pairs an Electron desktop interface with a Python command-line core and is aimed at macOS, Windows and Linux users who want to enlarge their own images locally rather than through a hosted service.

What it is

Final2x is published by EutropicAI and is written primarily in TypeScript. Its desktop application is assembled with Electron, Vue 3 and naive-ui, with electron-vite as the credited build reference, while the actual super-resolution work is handled by a Python backend built on PyTorch. The project's topic list places it squarely in the computer-vision and low-level-vision ecosystem, alongside image-processing and super-resolution concerns, and the interface ships with internationalisation support. Version 4.0.0 moved the backend onto cccv, the project's own backend library, and added support for custom models; version 3.0.0 added support for Nvidia 50-series GPUs. The Python CLI and the desktop backend now live together in the ./core directory and are released with the desktop app under the same version number, keeping the graphical and command-line paths on one track.

The concrete problem it solves is the plumbing around super-resolution models. Running a PyTorch upscaling model directly usually means installing Python dependencies, obtaining checkpoints, and driving each one through its own script or notebook, with no shared interface for swapping models or for working on a laptop. Final2x packages that workflow into a single desktop application with a graphical interface, and exposes the same engine as a Final2x-core command-line entry point, so a model can be tried through a GUI and later reused in a script without switching ecosystems. The custom-model support introduced in version 4.0.0, demonstrated by the companion cccv_demo_remote_model project, is the clearest expression of that goal: the tool is a host for models rather than a single fixed upscaler.

Key capabilities

• A single cross-platform desktop application for macOS, Windows and Linux, built with Electron and Vue 3 rather than three separate codebases.
• Version 4.0.0 runs on the cccv backend and supports custom models, with a remote-model demonstration repository showing the extension path.
• Version 3.0.0 added Nvidia 50-series GPU support.
• A Python CLI, installed as Final2x-core, with a -h flag for checking that the installation succeeded and for inspecting available options.
• The CLI and desktop backend share the ./core directory and ship under one version number, so they do not drift apart.
• Interface internationalisation is a listed project capability, alongside the computer-vision and image-processing topics.
• Distribution through GitHub Releases, with additional Windows installation through winget and scoop.

Who uses it and how

• Desktop users on macOS, Windows and Linux who want super-resolution applied through a graphical interface instead of a Python script.
• Windows users who prefer package management through winget or scoop, accepting that those channels may not carry the newest version.
• Linux users who already run Python 3.9 or newer and PyTorch 2.0 or newer, install Final2x-core with pip, and add the libomp5 and xdg-utils system packages.
• Experimenters who want to plug in their own super-resolution models rather than accept the built-in set, using the custom-model support in version 4.0.0.
• The repository has 7,332 stars, 525 forks and two open issues, indicating a widely watched project whose open-question queue is currently very short.

Getting started

Download the latest release from the GitHub Releases page, or use winget or scoop on Windows. macOS users must clear Gatekeeper on first run with sudo spctl --master-disable and xattr -cr /Applications/Final2x.app, while Linux users install pip install Final2x-core after confirming Python 3.9 or newer and PyTorch 2.0 or newer, plus apt install -y libomp5 xdg-utils.

How it compares

The facts provided do not name any comparable or competing tools, so Final2x stands alone in this registry. Its only named relationships are internal or infrastructural: the cccv backend it adopted in version 4.0.0, and the naive-ui and electron-vite projects credited as development references rather than alternatives.

When to use it — and when not to

A self-hoster takes on a Python environment of at least 3.9 with PyTorch 2.0 or newer, the libomp5 and xdg-utils system packages on Linux, and a Gatekeeper workaround on macOS before the application will launch. Anyone who does not want to manage a Python and PyTorch installation, or who needs a hosted service with no local dependencies, should look elsewhere, and the README documents GPU acceleration only through the Nvidia 50-series note. The licence is unambiguous under BSD-3-Clause, and the repository was last pushed in September 2026, but the README remains brief on hardware requirements and usage detail.

sudo spctl --master-disable
# Disable Gatekeeper, then allow applications downloaded from anywhere in System Preferences > Security & Privacy > General
xattr -cr /Applications/Final2x.app

pip install Final2x-core
Final2x-core -h # check if the installation is successful
apt install -y libomp5 xdg-utils