Video2X
C/C++ video upscaling and frame interpolation framework using Real-ESRGAN, Real-CUGAN, RIFE and Anime4K via Vulkan.
Real-ESRGAN, Real-CUGAN, Anime4K and RIFE through ncnn and Vulkan, with a Windows installer.
Video2X is a machine-learning video super-resolution and frame interpolation tool, rewritten in C/C++ for version 6. It suits people restoring old or low-resolution video and anime who have a Vulkan-capable GPU.
What it is
It runs in two modes: filtering (upscaling) and frame interpolation. Upscaling uses Anime4K v4 and custom MPV-compatible GLSL shaders, or Real-ESRGAN and Real-CUGAN via ncnn and Vulkan; interpolation uses RIFE. The README says it needs zero extra disk space during processing beyond the final output.
Who it's for
It suits archivists, anime fans and editors who want a free upscaler with a GUI on Windows or a container on Linux. It does not suit machines without a Vulkan GPU or older CPUs lacking AVX2, and macOS has no native build.
Key features
Two processing modes
Upscale video or interpolate extra frames with RIFE.
Several upscaling engines
Anime4K v4, Real-ESRGAN and Real-CUGAN are included.
Custom GLSL shaders
Accepts MPV-compatible GLSL shaders.
Windows Qt6 GUI
A desktop interface with an installer; GUI languages include English, Chinese and German.
Container image
Run it with Docker or Podman from the GitHub Container Registry.
No extra scratch space
Processes without writing large temporary frame files.
Up and running in 3 steps
Our honest verdict
What's great
- Windows installer with GUI, plus AppImage, AUR packages and container images for Linux
- Vulkan means NVIDIA, AMD and Intel GPUs are all supported
- Free Google Colab notebook lets you borrow a GPU if yours is weak
- Detailed documentation site covering building, installing and running
Where it falls short
- Needs a Vulkan-capable GPU and a CPU with AVX2 for the precompiled binaries
- Only Windows and Linux are supported; there is no macOS build
- AGPL-3.0 licence: modified versions offered over a network must share source
- Some documentation pages are still under construction, and the README's own video demos are labelled outdated
Pricing & license
Free under AGPL-3.0. The README mentions optional GitHub Sponsors and Patreon support for the developer, and free Google Colab use with Colab Pro available from Google; the only other cost is your own hardware.
Open remote AI-training roles
768 open remote AI-training roles, updated November 14, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Do I need an NVIDIA card?
No. The README requires a Vulkan-capable GPU from NVIDIA Kepler, AMD GCN 1.0 or Intel HD 4000 onwards.
Does it run on Mac?
The README lists Windows and Linux only, with containers for Linux and macOS via Docker or Podman.
Can I use it without a good GPU?
The README links a Google Colab notebook that borrows a GPU for free, with fair-use limits.
