Video encoder to assemble standard videos into 360° videos.
The challenge was to break YouTube copyright infringement by embedding classic videos into 360 ones and upload it on YouTube.
As it took a lot of time to process videos, we designed the executable with three parts to be able to distribute the computations on multiple computers.
The original video is split then each sub-video is sent to a different machine that will do the conversion process. Finally the sub-videos are reassembled into a whole.
To further increase performance we parallelized the conversions of the frames when a configurable option is activated.
This repository is the fruit of our research. In no event shall the authors of this project be liable for any use that would not respect the terms and conditions of YouTube's or any other website. We decline responsability for the use of our software for law infringement.
We made a talk about a possible application of our project for the "nuit de l'info" hosted by GConfs on the 12/08/2017 at EPITA.
The project is not installable, you however can get the executable (giving that you have the adequates prerequisites) by simply cloning the repository and running cmake.
- OpenCV 3.2.0
- Cmake 3.5.1
- ffmpeg 2.8.11
- libavfilter 5.40.101
- libavformat 56.40.101
- libavcodec 56.60.100
git clone https://github.com/Tauril/stealth-converter.git
cd stealth-converter
./configure -DPARALLEL=[0,1]
cmake .
make
The parallel option is to be set with the configure
./configure -DPARALLEL=[0,1]
help:
./stealth-converter [-h|--help]
split:
./stealth-converter [-s|--split] <Video_Path>
convert:
./stealth-converter [-c|--convert] <Video_Path>
reassemble:
./stealth-converter [-r|--recombine] <Sub_videos>
- Add option to choose the conversion face
- Add option to generate noise
- Gaël Gilet-Couraud
- Guillaume Marques - Tauril
- Loic Banet - banetl
This project is licensed under the MIT License - see the LICENSE.md file for details