Video2X
App : A product you use directly
Local video upscaling and frame interpolation (Real-ESRGAN, RIFE), open source.
The best free tool for restoring and enlarging a video.
Security reflects where data goes, requested permissions and incidents on record at the verification date. It is not a guarantee of zero risk: check the deployment mode and the connections you enable.
What it does, concretely
Old or low-resolution videos lose quality as soon as they are displayed on a large screen. Video2X lets you increase a video's resolution and smooth its motion, entirely on your own machine, without sending your files to a remote server.
The software offers two main modes: upscaling (image enlargement) and frame interpolation (increasing the number of frames per second). It relies on several well-established AI engines: Real-ESRGAN, Real-CUGAN, Anime4K v4, and RIFE, all running via ncnn and Vulkan to make use of the GPU.
Video2X is aimed at video enthusiasts, anime fans, and anyone wishing to restore old recordings without a subscription or internet connection. A graphical interface is available on Windows, and a Docker image simplifies installation on Linux and macOS.
However, the software requires a minimum level of technical knowledge for installation on Linux or macOS, and needs Vulkan-compatible hardware. Users looking for a fully turnkey solution with no configuration will be better served by an online tool.
What it offers
- ·AI-based video upscaling with Real-ESRGAN, Real-CUGAN, and Anime4K v4.
- ·Frame interpolation with RIFE to increase video smoothness (for example from 30 to 60 frames per second).
- ·Support for custom MPV-compatible GLSL shaders for advanced users.
- ·Native graphical interface on Windows with a one-click installer, available in French.
- ·Fully local processing: no video data is sent to an external server.
- ·Available as a Docker image for simplified deployment on Linux and macOS.
- ·Free use on Google Colab for machines without a powerful GPU, borrowing an NVIDIA T4, L4, or A100 GPU for up to 12 hours per session.
- ·No additional disk space required during processing, only for the final output file.
In practice
- Price
- Free
- Commercial use
- Commercial use: yes
- Available on
- windows, linux, macos
- Licence
- AGPL-3.0
- Runs locally
- Yes, 4 GB VRAM recommended
- FR interface
- No
Security and data
- Rating
- Security: nothing to flag
- Your data
- Data stays local
- Maintenance
- Inactive for over six months
- Last release
- 24 January 2025
Links and repository checked on 25 October 2026 · added on 1 January 2018