Final2x v4.0: A Cross‑Platform Image Super‑Resolution Tool

What is Final2x?

Final2x is a cross‑platform desktop application that brings cutting‑edge image super‑resolution to everyday users. At its core, the project relies on Final2x‑core, a PyTorch‑based library that implements a variety of state‑of‑the‑art models. The GUI layer is built with Electron + Vue 3 + TypeScript, making the app lightweight, responsive, and easy to maintain.

Why you should care

• GPU acceleration – Final2x now supports Nvidia 50‑series GPUs in addition to the 40‑series, giving you lightning‑fast upscaling on modern hardware.
• Custom model support – You can drop in your own PyTorch model via a simple config file, or use any of the bundled pretrained models.
• Cross‑platform – Build and run the same binary on Windows, macOS, or Linux with the same user experience.
• Open source – MIT‑style BSD‑3 license means you can use, modify, and distribute the code freely.

Getting Started

1. Download the Release

The easiest way to try Final2x is to download the latest binary from GitHub:

• Windows: Final2x‑v4.0.0-win-x64.exe
• macOS: Final2x‑v4.0.0-darwin-x64.zip
• Linux: Final2x‑v4.0.0-linux-x86_64.AppImage

Alternatively, use a package manager:

# Windows (winget)
winget install Final2x
# macOS (Homebrew)
brew install --cask final-2x

2. Install on macOS

1. Open the downloaded .zip.
2. Drag the Final2x.app to Applications.
3. Run the app once; you may need to allow it through Gatekeeper.

sudo spctl --master-disable && xattr -cr /Applications/Final2x.app

3. Install on Linux

Linux users can install the Python package for the backend and run the app from the command line:

# Install dependencies
sudo apt install -y libomp5 xdg-utils
# Install the core library
pip install Final2x-core
# Launch the GUI (optional)
final2x-gui

You can also run the standalone AppImage directly:

chmod +x Final2x‑v4.0.0-linux-x86_64.AppImage
./Final2x‑v4.0.0-linux-x86_64.AppImage

4. Using Final2x

• Open an image – Drag‑and‑drop a JPEG, PNG, or BMP into the main window.
• Choose scale – Set 2×, 4×, or 8× upscaling factors.
• Select model – Pick a pretrained model from the drop‑down or add a custom one via config.json.
• Adjust options – Fine‑tune denoise, clip space, and color preservation settings.
• Export – Save the upscaled image to any directory.

5. Custom Models

Final2x makes it straightforward to plug in your own model. Add a YAML entry in ~/.config/Final2x/config.yaml:

custom_models:
  - name: my_unet
    path: /path/to/unet.pt
    width: 256
    height: 256

Then select my_unet from the GUI. The framework automatically handles model loading and inference.

Contributing

Fork the repo, create a feature branch, and submit a pull request. The project uses pnpm and vitest for testing. Please run pnpm run test before submission.

License

Final2x is released under the BSD 3‑Clause license. Refer to the LICENSE file in the repository for full details.

In a Nutshell

Final2x v4.0 offers a polished, GPU‑accelerated super‑resolution experience on any desktop. With a modular core, easy custom model integration, and a modern Electron UI, it's an ideal showcase for practitioners and hobbyists alike. Grab the binary today or build from source and start making your images look sharper than ever.