Video2X 6.0.0 highlights:

• Complete rewrite of the Video2X project in C/C++.
• Faster and more efficient architecture.
• Cross-platform support for Windows and Linux.
• Vastly improved output quality.
• New GUI and installer for easy setup on Windows.

Your system must meet the minimum hardware requirements below to run Video2X.

• CPU
  • The precompiled binaries require CPUs with AVX2 support.
  • Intel: Haswell (Q2 2013) or newer
  • AMD: Excavator (Q2 2015) or newer
• GPU
  • The GPU must support Vulkan.
  • NVIDIA: Kepler (GTX 600 series, Q2 2012) or newer
  • AMD: GCN 1.0 (Radeon HD 7000 series, Q1 2012) or newer
  • Intel: HD Graphics 4000 (Q2 2012) or newer

Download the Latest Windows Installer Executable (6.4.0)

You can download the latest Windows release on the releases page. For basic GUI usage, refer to the documentation. If you're unable to download directly from GitHub, try the mirror site. The GUI currently supports the following languages:

• English (United States)
• 简体中文（中国）
• 日本語（日本）
• Português (Portugal)
• Français (France)
• Deutsch (Deutschland)

Video2X packages are available for the Linux distros listed below. A universal AppImage is also available for other distros. If you'd like to build it from source code, refer to the PKGBUILD file for a general overview of the required dependencies and commands.

• Arch Linux: AUR packages, maintained by @K4YT3X.
  • aur/video2x
  • aur/video2x-git
  • aur/video2x-qt6
  • aur/video2x-qt6-git
• Arch Linux (Chinese Mainland): archlinuxcn packages, maintained by @Integral-Tech.
  • archlinuxcn/video2x
  • archlinuxcn/video2x-git
  • archlinuxcn/video2x-qt6
  • archlinuxcn/video2x-qt6-git
• Other distros: Video2X-x86_64.AppImage on the releases page.

This fork currently supports source builds on the latest macOS running on Apple Silicon. The canonical contributor workflow is the thin just entrypoint backed by shared CMake presets, so contributors do not need to reconstruct raw cmake -D... invocations by hand.

Before building, make sure the prerequisite surface is in place:

• Xcode or the Apple Command Line Tools selected through xcode-select
• Homebrew under /opt/homebrew
• cmake, ninja, ffmpeg, and pkg-config (Homebrew's pkgconf formula provides it)
• Vulkan portability tooling: either a Vulkan SDK or the Homebrew MoltenVK stack, plus glslangValidator
• Extra system-mode packages for the validated macos-system-* path: Homebrew ncnn, spdlog, and boost

From the repo root, preflight those requirements and then run one configure/build pair:

just doctor-macos
just configure-macos-system-release
just build-macos-system-release

The matching Debug path is:

just doctor-macos
just configure-macos-system-debug
just build-macos-system-debug

Use the vendored configure-macos-vendored-* and build-macos-vendored-* recipes only when those system-only Homebrew packages are unavailable. Vendored mode still expects a populated third_party/ checkout; if configure stops inside third_party/boost, refresh the local checkout state before assuming the preset names are wrong. All of these just recipes remain thin wrappers around the shared presets in CMakePresets.json.

After the canonical Release build, validate the macOS runtime surface from the repo root with:

just smoke-macos
just list-devices-macos
just sample-macos-realesrgan

These Phase 3 commands target the built build/macos-system-release/video2x binary by default. If you need to probe the install tree as a secondary check, run the script directly with --binary-mode installed, for example:

./scripts/macos_runtime_validation.sh smoke --binary-mode installed

just list-devices-macos is expected to pass only when the built binary exists locally and the macOS Vulkan portability stack is working end to end: MoltenVK or equivalent tooling is installed, the loader advertises VK_KHR_portability_enumeration, and the command prints at least one detected GPU name before exiting 0. just sample-macos-realesrgan is the canonical sample workload because it uses the validated ncnn-backed Real-ESRGAN path, generates a short local clip with ffmpeg when you do not supply one, and confirms the output video with ffprobe. The repo does not assume data/standard-test.mp4 exists for this proof flow, and libplacebo remains optional rather than the canonical macOS validation target until it is revalidated locally.

For the contributor-facing workflow surface, see CONTRIBUTING.md. The docs book now includes the first-class macOS build guide and the initial macOS GPU strategy, which document the same Apple Silicon command surface and the Vulkan SDK plus MoltenVK path this fork currently treats as canonical.

Video2X container images are available on the GitHub Container Registry for easy deployment on Linux and macOS. If you already have Docker/Podman installed, only one command is needed to start upscaling a video. For more information on how to use Video2X's Docker image, please refer to the documentation.

You can use Video2X on Google Colab for free if you don't have a powerful GPU of your own. You can borrow a powerful GPU (NVIDIA T4, L4, or A100) on Google's server for free for a maximum of 12 hours per session. Please use the free resource fairly and do not create sessions back-to-back and run upscaling 24/7. This might result in you getting banned. You can get Colab Pro/Pro+ if you'd like to use better GPUs and get longer runtimes. Usage instructions are embedded in the Colab Notebook.

Join our Telegram discussion group to ask any questions you have about Video2X, chat directly with the developers, or discuss super resolution, frame interpolation technologies, or the future of Video2X in general.

Comprehensive documentation for Video2X is available at https://docs.video2x.org/. It offers detailed instructions on how to build, install, use, and develop with this program.

Upscale demo: Spirited Away's movie trailer

• Spirited Away: YouTube | Bilibili
  • 360P to 4K
  • The original video's copyright belongs to 株式会社スタジオジブリ
• Bad Apple!!: YouTube | Bilibili
  • 384P 30 FPS to 4K 120 FPS with waifu2x and DAIN
  • The original video's copyright belongs to あにら
• The Pet Girl of Sakurasou: YouTube | Bilibili
  • 240P 29.97 to 1080P 60 FPS with waifu2x and DAIN
  • The original video's copyright belongs to ASCII Media Works

The following clip can be used to test if your setup works properly. This is also the standard clip used for running performance benchmarks.

For the macOS Phase 3 validation flow above, downloading this clip is optional. The canonical sample-macos-realesrgan command generates a short local input with ffmpeg unless you pass an explicit input file to scripts/macos_runtime_validation.sh.

• Standard Test Clip (240P) 4.54 MiB
• Real-CUGAN Upscaled Sample (1704P) 3.5 MiB
• Real-ESRGAN Upscaled Sample (1704P) 3.1 MiB
• waifu2x Upscaled Sample (1080P) 4.54 MiB
• Ground Truth (1080P) 22.2 MiB

The original clip came from the anime "さくら荘のペットな彼女."
Copyright of this clip belongs to 株式会社アニプレックス.

This project is licensed under GNU AGPL version 3.
Copyright (C) 2018-2025 K4YT3X and contributors.

This project includes or depends on these following projects:

More licensing information can be found in the NOTICE file.

Special thanks to the following individuals for their significant contributions to the project, listed in alphabetical order.

• @ArchieMeng
• @BrianPetkovsek
• @Integral-Tech
• @ddouglas87
• @lhanjian
• @nihui
• @sat3ll