Linux 需先装 Python >= 3.10 和 PyTorch >= 2.0，并额外安装 libomp5、xdg-utils 等系统库；macOS 首次启动可能被拦截，需执行 xattr -cr /Applications/Final2x.app；winget/Scoop 等包管理器版本可能落后于最新 release。
Final2x
EutropicAIBSD-3-Clause活跃维护跨平台图片超分辨率工具，基于 PyTorch 模型把低清图片放大变清晰，支持多语言，桌面端开箱即用。
a cross-platform image super-resolution tool
v4.1.0发布于 2026-09-30 16:57 UTC7,406star +6 / 7 天527fork同步于 2026-11-10 11:32 UTC
pip install Final2x-core
Final2x-core -h
sudo apt install -y libomp5 xdg-utils 硬件要求按 README 说明
README 未说明显存或 GPU 型号要求；Linux 端需安装 PyTorch >= 2.0，未提及是否可纯 CPU 运行。
跨平台图像超分辨率工具，内置多种模型也可加载自定义模型放大图片
跨平台图片超分辨率工具，基于 PyTorch 模型把低清图片放大变清晰，支持多语言，桌面端开箱即用。
核心能力
- 支持 Windows、macOS、Linux 三平台桌面应用
- 内置超分模型，也可加载自定义模型
- 提供 Python CLI 与桌面端共用同一后端
- 基于 cccv 图像修复与超分后端
容易踩的坑
内容由模型根据 README 生成，可能不完整或不准确，以官方文档为准。看原文 README
- v4.1.0最新正式版2026-09-30 · 2026-09-30 16:57 UTC
What's New
- Add FP32 / FP16 / BF16 precision selection.
- Improve the settings page layout.
Bug Fix
- Fix HTTPS model downloads failing with
CERTIFICATE_VERIFY_FAILEDon clean macOS installations without Homebrew. Model downloads now use macOS system certificate trust. (#615, #716) - Fix repeated IPC handler registration.
- Improve error handling when the backend executable is missing.
- Update dependencies.
Maintenance
- Merge Final2x-core into the main repository and align desktop and core versions.
- Refactor the build and release workflows.
- Update documentation.
What's Changed
- build(deps): bump systeminformation from 5.27.10 to 5.30.8 by @dependabot[bot] in https://github.com/EutropicAI/Final2x/pull/691
- build(deps-dev): bump vite from 7.1.9 to 7.1.11 by @dependabot[bot] in https://github.com/EutropicAI/Final2x/pull/686
- build(deps-dev): bump vitest from 3.2.4 to 3.2.6 by @dependabot[bot] in https://github.com/EutropicAI/Final2x/pull/700
- refactor: merge Final2x-core into the Final2x repo by @NULL204 in https://github.com/EutropicAI/Final2x/pull/704
- Del docs by @CHbloom in https://github.com/EutropicAI/Final2x/pull/707
- refactor: build pipeline and CI by @NULL204 in https://github.com/EutropicAI/Final2x/pull/708
- refactor: streamline core path retrieval and command execution by @NULL204 in https://github.com/EutropicAI/Final2x/pull/709
- fix: restore core CI and report backend launch failures by @NULL204 in https://github.com/EutropicAI/Final2x/pull/712
- fix: macOS certificate trust and IPC reactivation by @NULL204 in https://github.com/EutropicAI/Final2x/pull/713
- feat: add FP32/FP16/BF16 precision selection by @NULL204 in https://github.com/EutropicAI/Final2x/pull/714
- fix: replace the External Link by @NULL204 in https://github.com/EutropicAI/Final2x/pull/715
- ci: adopt OIDC, replace issue helper by @NULL204 in https://github.com/EutropicAI/Final2x/pull/717
- ci: skip PyPI on forks to allow GitHub releases and update Docs by @NULL204 in https://github.com/EutropicAI/Final2x/pull/718
- chore: move auto-open output toggle into settings rows by @NULL204 in https://github.com/EutropicAI/Final2x/pull/720
- fix: use macOS system trust for core downloads by @NULL204 in https://github.com/EutropicAI/Final2x/pull/716
- release: v4.1.0 by @NULL204 in https://github.com/EutropicAI/Final2x/pull/722
New Contributors
- @CHbloom made their first contribution in https://github.com/EutropicAI/Final2x/pull/707
Full Changelog: https://github.com/EutropicAI/Final2x/compare/v4.0.0...v4.1.0
- v4.0.02025-10-05 · 2025-10-05 06:15 UTC
查看更新说明
- Final2x-core v4.0.0
What's New
- New Backend: now we use the cccv backend, supporting custom models and more. See https://github.com/EutropicAI/cccv_demo_remote_model . Thanks to @yuygfgg for testing on macOS~
You can load a custom model from git url (git cli installed)
Or load from local path
Bug Fix
- Dependency bumps to latest
- Cleaner exit
- Refactor some code...
- 2025-08-16v3.0.02025-08-16 · 2025-08-16 13:38 UTC
查看更新说明
- Final2x-core: v3.2.0
Final2x v3.0.0 is now available!
The most important update: we’ve got a new icon — it’s absolutely crucial and instantly boosts AI performance!
New Features
- Tile processing: optional switch to enable/disable tile inference
- Save format: user-selectable output image format
Breaking Changes
- Bumped PyTorch to 2.7.1+cu128 for NVIDIA 50-series support; older GPUs (10s) may work not well
- MacOS x64 is no longer included in official pre-built packages
Bug Fix
- Update dependencies
- Fix some other issues
- 2024-12-14v2.1.02024-12-14 · 2024-12-14 21:45 UTC
- 2024-11-09v2.0.02024-11-09 · 2024-11-09 14:07 UTC
查看更新说明
Final2x v2.0.0 has been released!
- Final2x-core: v3.0.1
- ccrestoration: v0.1.2
We are thrilled to announce the release of Final2x v2.0.0, which brings a significant upgrade to our image restoration capabilities. This release marks a major milestone as we transition to utilizing ccrestoration (PyTorch) for our algorithm implementation, enhancing performance and expanding the range of supported models.
Breaking Changes
- Windows User Changes: Only Nvidia users can use GPU acceleration.
- Linux User Changes: Linux users need to install dependencies using pip, as we will no longer provide pre-built Final2x-core. (Install ROCm PyTorch for AMD GPU)
- MacOS Intel User Changes: MacOS Intel users can not use GPU acceleration, only CPU.
Features
- Expanded Model Support: We are excited to introduce support for a variety of new models, including RealESRGAN, RealCUGAN, EDSR, SwinIR, SCUNet, DAT......
- Future-Proofing: We will continue to add support for more models in future updates!
用户点评
登录用户的真实使用记录，可按版本和硬件筛选共 0 条
还没有人点评。第一个说出真实体验的人最有价值。
A cross-platform image super-resolution tool for Windows, macOS, and Linux.
Use built-in models or load your own. See the custom model demo to get started.
🖼️ Screenshots
📦 Installation
Download the latest release for your platform.
🪟 Windows
Install from the release page, or use a package manager such as winget or Scoop. Package manager versions may lag behind the latest release.
🍎 macOS
If macOS blocks the app on first launch, run the following command in Terminal after placing it in /Applications:
xattr -cr /Applications/Final2x.app
🐧 Linux
Install Python >= 3.10 and PyTorch >= 2.0, then install the backend and required system libraries. On Debian or Ubuntu:
pip install Final2x-core
Final2x-core -h # Verify the backend installation
sudo apt install -y libomp5 xdg-utils
🧩 References
The Python CLI and desktop backend live in core and share the desktop app's version number.
This project builds on the following open-source projects:
- cccv — image restoration and super-resolution backend
- naive-ui
- electron-vite
📄 License
This project is licensed under the BSD 3-Clause - see the LICENSE file for details.
💙 Acknowledgements
Feel free to reach out to the project maintainers with any questions or concerns~
opencv
opencv·C++
开源计算机视觉库，提供图像处理、视频分析和深度学习推理的 C++ 接口，是视觉应用开发的基础设施。
计算机视觉+4图像处理
91525+4157k5.1.02026-11-10 12:05 UTC
upscayl
upscayl·TypeScript
跨平台的开源 AI 图片放大工具，基于 ESRGAN 等模型把低分辨率图片无损放大并增强细节，有图形界面，普通用户也能直接上手。
图像超分+4图像处理
52145+2652.7kv2.15.02026-10-05 19:13 UTC
supervision
roboflow·Python
一套可复用的计算机视觉工具库，提供目标检测、实例分割、分类等任务的标注处理、指标计算与可视化组件，帮开发者省去重复造轮子的时间。
计算机视觉+4图像处理
51577+444.9k0.31.52026-11-10 11:15 UTC
sharp
lovell·JavaScript
基于 libvips 的高性能 Node.js 图像处理库，能快速完成 JPEG、PNG、WebP、AVIF、TIFF 的缩放裁剪与格式转换，适合服务端批量处理图片。
图像处理+4图像处理
32836+71.4kv0.35.52026-11-09 13:05 UTC