Video2X logo

Video2X

Machine learning video super resolution and frame interpolation for Windows, Linux, and containerized use

Repository activity
  • Stars20.3k
  • Forks1.8k
  • Open Issues113
License

AGPL-3.0

Languages
  • C++
  • CMake
  • Just
Video2X screenshot

About Video2X

Video2X is a video super-resolution and frame-interpolation framework for improving video quality. It upscales footage to a higher resolution and raises frame rate, with a GUI and installer for Windows alongside command-line and packaged builds for Linux.

It runs in two modes: filtering for upscaling and frame interpolation for smoother motion. Output can be driven by Anime4K v4, Real-ESRGAN, Real-CUGAN, and RIFE, including custom mpv-compatible GLSL shaders. A C and C++ processing pipeline keeps it fast and uses no extra disk space beyond the final output.

Video2X runs on Windows and Linux, and can run in Docker or Podman containers for headless and server use. It is licensed under the GNU AGPL v3 and needs a Vulkan-capable GPU.

Key features

  • AI video upscaling to higher resolution
  • Frame interpolation for smoother, higher-FPS video
  • Anime4K v4, Real-ESRGAN, Real-CUGAN, and RIFE models
  • Custom mpv-compatible GLSL shaders
  • Runs in Docker or Podman containers

Details

On GitHub since
2018
Platforms
Windows · Linux · Docker
Deployment
Self-hostable · docker
License
GNU AGPL v3
Modes
Filtering · frame interpolation
GPU
Vulkan required