Video2x

Video2X is an open source, machine learning based framework for video super resolution and frame interpolation. It upscales video to a higher resolution and can raise the frame rate by generating intermediate frames, working from the original footage rather than requiring manual restoration. The project's documentation site covers installation and usage, and the project itself is hosted on GitHub. It is aimed at people who want to enhance older or low resolution footage, including anime, on their own hardware. Version 6 is a complete rewrite in C and C++ with support for Windows and Linux. It offers two modes, filtering for upscaling and frame interpolation. Upscaling uses Anime4K version 4 and custom MPV compatible GLSL shaders, as well as the Real ESRGAN and Real CUGAN models, while interpolation uses RIFE, all running through ncnn and Vulkan. The pipeline needs no extra disk space during processing beyond the final output. On Windows a graphical interface and installer are provided, with the interface available in English, Simplified Chinese, Japanese, Portuguese, French and German. Linux users can install packages for Arch Linux, use a universal AppImage, or run a container image with Docker or Podman. The hardware must support Vulkan, and the precompiled binaries need a CPU with AVX2 support. The project is free to download. The documentation site publishes no pricing page, and the project accepts donations through sponsorship platforms.