Final2x
github.com
Spec sheet
- Purpose
- Final2x is a cross-platform image super-resolution tool for Windows, macOS, and Linux.Read 12 Oct 2026 · github.com
- Support
- The README invites users to contact the project maintainers with questions or concerns.Read 12 Oct 2026 · github.com
- License
- The repository uses the BSD 3-Clause License, which permits redistribution and use subject to its listed conditions.Read 12 Oct 2026 · github.com
- Backend
- Final2x-core provides the image super-resolution command-line interface and desktop backend.Read 12 Oct 2026 · github.com
- CLI configuration
- The CLI accepts configuration as Base64, JSON, or YAML, and can save logs.Read 12 Oct 2026 · github.com
- CLI installation
- The CLI can be installed from PyPI with pip install Final2x-core.Read 12 Oct 2026 · github.com
- Custom models
- The v4.0.0 release says custom models can be loaded from a Git URL or a local path.Read 12 Oct 2026 · github.com
- Desktop app
- YesRead 22 Sep 2026 · github.com
- Downloads
- The README directs users to download the latest release for their platform.Read 12 Oct 2026 · github.com
- GPU limitations
- The v2.0.0 release notes say Windows GPU acceleration is for Nvidia users, while Intel macOS users can use only the CPU.Read 12 Oct 2026 · github.com
- GPU support
- The v2.0.0 release notes say Windows GPU acceleration is available only to Nvidia users and macOS Intel users can use CPU only.Read 12 Oct 2026 · github.com
- Linux requirements
- The Linux installation instructions require Python 3.10 or newer and PyTorch 2.0 or newer, plus system libraries on Debian or Ubuntu.Read 12 Oct 2026 · github.com
- macOS launch
- The README says users may need to run xattr -cr /Applications/Final2x.app if macOS blocks the app on first launch.Read 12 Oct 2026 · github.com
- Model support
- The v2.0.0 release lists RealESRGAN, RealCUGAN, EDSR, SwinIR, SCUNet, and DAT among its supported models.Read 12 Oct 2026 · github.com
- Models
- The project says users can use built-in models or load their own.Read 12 Oct 2026 · github.com
- Output options
- The v3.0.0 release added optional tile processing and user-selectable output image formats.Read 12 Oct 2026 · github.com
- Output settings
- A release added a user-selectable output image format and an optional tile-inference switch.Read 12 Oct 2026 · github.com
- Precision
- Version 4.1.0 added FP32, FP16, and BF16 precision selection.Read 12 Oct 2026 · github.com
- Release verification
- The v4.0.0 release page says its commit was created with GitHub’s verified signature.Read 12 Oct 2026 · github.com
- Upscaling method
- aiRead 22 Sep 2026 · github.com
- Windows installation
- The README says Windows users can install from the release page or use winget or Scoop, whose packages may lag behind the latest release.Read 12 Oct 2026 · github.com
Where Specifiction places it
- Image Upscaling Software#93 of 997.0