video2x/README.md

Video2X Lossless Video Enlarger

Official Discussion Group (Telegram): https://t.me/video2x

Prerequisites

Component names that are bolded are mandatory.

Component names that are italicized can be automatically downloaded and configured with the video2x_setup.py script.

1. Operating System: Windows
2. AMD GPU / Nvidia GPU
3. AMD GPU driver / Nvidia GPU driver / Nvidia CUDNN
4. FFMPEG
5. waifu2x-caffe / waifu2x-converter-cpp

Recent Changes

2.7.0 (March 30, 2019)

• Added support for different extracted image formats.
• Redesigned FFMPEG wrapper, FFMPEG settings are now customizable in the video2x.json config file.
• Other minor enhancements and adjustments (e.g. argument -> method variable)

Setup Script 1.2.0 (March 26, 2019)

• video2x_setup.py script can now automatically download and configure waifu2x-converter-cpp.
• replaced old progress indicator with progress bar.

2.6.3 (March 24, 2019)

• Added image cleaner by @BrianPetkovsek which removes upscaled frames.
• Fixed some PEP8 issues.
• Exceptions in waifu2x are now caught, and script will now stop on waifu2x error instead of keep going on to FFMPEG.

Description

Video2X is an automation software based on waifu2x image enlarging engine. It extracts frames from a video, enlarge it by a number of times without losing any details or quality, keeping lines smooth and edges sharp.

For short: Video2X enlarges your video without losing details

Watch for the sharper edges in this screenshot around the shadows:

You can also watch the YouTube video Demo: https://www.youtube.com/watch?v=PG94iPoeoZk

Clip is from trailer of animated movie "千と千尋の神隠し". Copyright belongs to "株式会社スタジオジブリ (STUDIO GHIBLI INC.)". Will delete immediately if use of clip is in violation of copyright.

Screenshot

---

Documentations

Video2X Wiki

You can find all detailed user-facing and developer-facing documentations in the Video2X Wiki. It covers everything from step-by-step instructions for beginners, to the code structure of this program for advanced users and developers. If this README page doesn't answer all your questions, the wiki page is where you should head to.

Step-By-Step Tutorial (Nvidia GPUs)

For those who want a detailed walk-through of how to use Video2X, you can head to the Step-By-Step Tutorial wiki page. It includes almost every step you need to perform in order to enlarge your first video.

Waifu2X Drivers

Go to the Waifu2X Drivers wiki page if you want to see a detailed description on the different types of waifu2x drivers implemented by Video2X. This wiki page contains detailed difference between different drivers, and how to download and set each of them up for Video2X.

---

Quick Start

Prerequisites

• Python 3
  Download: https://www.python.org/downloads/windows/
• FFMPEG Windows Build
  Download: https://ffmpeg.org/download.html
• waifu2x-caffe (for Nvidia CUDA/CUDNN)
  Download: https://github.com/lltcggie/waifu2x-caffe/releases
• waifu2x-converter-cpp (required for AMD, OpenCL and OpenGL processing)
  Download: https://github.com/DeadSix27/waifu2x-converter-cpp/releases

Installing Dependencies

First, clone the video2x repository.

git clone https://github.com/K4YT3X/video2x.git
cd video2x/bin

Then you may run the video2x_setup.py script to install and configure the dependencies automatically. This script is designed and tested on Windows 10.

This script will install the newest version of ffmpeg, either or both of waifu2x-caffe and waifu2x-converter-cpp to %LOCALAPPDATA%\\video2x and all required python libraries.

python video2x_setup.py

Alternatively, you can also install the dependencies manually. Please refer to the prerequisites section to see what's needed.

Then you'll need to install python dependencies before start using video2x. Install simply by executing the following command.

pip install -r requirements.txt

Note that all command line arguments/options overwrite configuration file settings.

Sample Videos

If you can't find a video clip to begin with, or if you want to see a before-after comparison, we have prepared some sample clips for you. The quick start guide down below will also be based on the name of the sample clips.

• Sample Video Original (240P) 1.7MB
• Sample Video Upscaled (1080P) 4.8MB

Clip is from anime "さくら荘のペットな彼女". Copyright belongs to "株式会社アニプレックス (Aniplex Inc.)". Will delete immediately if use of clip is in violation of copyright.

Nvidia CUDA (waifu2x-caffe)

Enlarge the video to 1920x1080 using CUDA. You may also use the -r/--ratio option.

python video2x.py -i sample_input.mp4 -o sample_output.mp4 -m gpu --width=1920 --height=1080

Nvidia CNDNN

Enlarge the video to 1920x1080 using CUDNN. You may also use the -r/--ratio option.

python video2x.py -i sample_input.mp4 -o sample_output.mp4 -m cudnn --width=1920 --height=1080

AMD or Nvidia (waifu2x-converter-cpp OpenCL)

Enlarge the video by 2 times using OpenCL. Note that waifu2x-converter-cpp doesn't support width and height. You'll also have to explicitly specify that the driver to be used is waifu2x_converter.

python video2x.py -i sample_input.mp4 -o sample_output.mp4 -m gpu -r 2 -d waifu2x_converter

CPU

Enlarge the video to 1920x1080 using the CPU. You may also use the -r/--ratio option. This is potentially much slower than using a GPU. The configuration file for this method is similar to the previous methods.

python video2x.py -i sample_input.mp4 -o sample_output.mp4 -m cpu --width=1920 --height=1080

---

Full Usage

General Options

-h, --help

show this help message and exit

File Options

-i INPUT, --input INPUT

Source video file/directory (default: None)

-o OUTPUT, --output OUTPUT

Output video file/directory (default: None)

Upscaling Options

-m {cpu,gpu,cudnn}, --method {cpu,gpu,cudnn}

Upscaling method (default: gpu)

-d {waifu2x_caffe,waifu2x_converter}, --driver {waifu2x_caffe,waifu2x_converter}

Waifu2x driver (default: waifu2x_caffe)

-y MODEL_DIR, --model_dir MODEL_DIR

Folder containing model JSON files

-t THREADS, --threads THREADS

Number of threads to use for upscaling (default: 5)

-c CONFIG, --config CONFIG

Video2X config file location (default: video2x\bin\video2x.json)

-b, --batch

Enable batch mode (select all default values to questions)

Scaling Options

--width WIDTH

Output video width

--height HEIGHT

Output video height

-r RATIO, --ratio RATIO

Scaling ratio

---

License

Licensed under the GNU General Public License Version 3 (GNU GPL v3) https://www.gnu.org/licenses/gpl-3.0.txt

(C) 2018-2019 K4YT3X

Credits

This project relies on the following software and projects.

• FFMPEG
• waifu2x-caffe
• waifu2x-converter-cpp

Related Resources

• Dandere2x: Dandere2x is a lossy video upscaler also built around waifu2x, but with video compression techniques to shorten the time needed to process a video.