DISCLAIMER:

  1. We do not have any rights to the video content.
  2. All video content must be selected from sources of your choice.
  3. The MPQUIC-SBD repositories provide video scripts only for processing user-provided video content by encoding it and splitting it into segments of varying quality levels.

Notes:

  • Example of video source: Freely licensed video content can be found in public domain repositories such as Wikimedia.
  • Warning on video preparation: This is an offline process that may take a significant amount of time, depending on the video’s length and the desired quality. We recommend performing this process outside the Vagrant VM, e.g., directly on your host.
  • Information on reproducibility: In our experiments (refer to our publications), we use a 4K (3840 x 2160), 60 fps video file, 50 minutes in length, which is encoded in 11 different qualities and split into 2-, 4-, and 10-second segments.

Offline processing of video segments

Step 1: Requirements

First, install ffmpeg and GPAC’s MP4Box on your host system.

Run on your host:

1

sudo apt-get install -y ffmpeg gpac

Install ffmpeg and gpac using your package manager, as shown in the commands above. Just for reference, the versions we used were:

  • Ubuntu 18.04.6 LTS
  • gcc 6.5.0
  • MP4Box 0.5.2-426-gc5ad4e4+dfsg5-3ubuntu0.1
  • ffmpeg 7:3.4.11-0ubuntu0.1

Then, navigate to the mpquic-sbd-vagrant/Workspace folder and download our video processing scripts from our repository by running:

1
2

cd ~/mpquic-sbd-vagrant/Workspace
git clone https://github.com/mpquic-sbd/video.git

Navigate to the mpquic-sbd-vagrant/Workspace/video folder, where you will find the following Python scripts:

1
2
3
4
5

~/mpquic-sbd-vagrant/Workspace/video
├── video_download_process.py # downloads (wget) and processes (ffmpeg) a video file
├── video_sara_mpd.py         # creates an MPD file for the SARA ABR format
├── video_split.py            # splits the processed videos into the desired segment length (in seconds)
├── video_utils.py            # set of functions used to download, process, and split video files

Step 2: Download and process the video

The script video_download_process.py downloads and processes a video file (WEBM or MP4). Using ffmpeg, the original video is encoded into 11 different quality levels, as done in our experiments (refer to our publications).

In the mpquic-sbd-vagrant/Workspace/video folder, download and process the video by running:

1
2

cd ~/mpquic-sbd-vagrant/Workspace/video
python3 video_download_process.py <video file URL>

For example, the Wikimedia’s video File:Walking in LONDON - England (UK) - Westminster to Piccadilly Circus (2019) - 4K 60fps (UHD).webm is a 4K, 60fps, 50+ minute video, which can be easily downloaded and processed as follows (note that this process may take several hours to complete):

1
2

cd ~/mpquic-sbd-vagrant/Workspace/video
python3 video_download_process.py https://upload.wikimedia.org/wikipedia/commons/b/bf/Walking_in_LONDON_-_England_%28UK%29_-_Westminster_to_Piccadilly_Circus_%282019%29_-_4K_60fps_%28UHD%29.webm

The video_download_process.py script is expected to produce the following output files:

1
2
3
4
5
6
7
8
9
10
11
12
13

~/mpquic-sbd-vagrant/Workspace/video
├── downloaded_video.webm   # original downloaded video file
├── video_21602p_60_fps.mp4 # q1:  processed video with 60 fps, 3840x2160  
├── video_2160p_30_fps.mp4  # q2:  processed video with 30 fps, 3840x2160 
├── video_14402p_60_fps.mp4 # q3:  processed video with 60 fps, 2560x1440
├── video_1440p_30_fps.mp4  # q4:  processed video with 30 fps, 2560x1440
├── video_10802p_60_fps.mp4 # q5:  processed video with 60 fps, 1920x1080
├── video_1080p_25_fps.mp4  # q6:  processed video with 25 fps, 1920x1080 
├── video_7202p_60_fps.mp4  # q7:  processed video with 60 fps, 1280x720   
├── video_720p_25_fps.mp4   # q8:  processed video with 25 fps, 1280x720  
├── video_480p_25_fps.mp4   # q9:  processed video with 25 fps, 854x480
├── video_360p_25_fps.mp4   # q10: processed video with 25 fps, 640x360
├── video_240p_25_fps.mp4   # q11: processed video with 25 fps, 426x240

Warning: do not change the file names, as it causes the other scripts to fail.

Step 3: Split the processed video into segments

The script video_split.py uses MP4Box to split the entire video file into chunks, i.e., segments in sequence, each containing a few seconds of the video. In our experiments (refer to our publications), we split the video into 2-, 4-, and 10-second segments. To do this, run the following commands:

1
2
3
4

cd ~/mpquic-sbd-vagrant/Workspace/video
python3  video_split.py  2
python3  video_split.py  4
python3  video_split.py  10

In addition to the original downloaded video (downloaded_video.webm) and the processed videos in 11 different qualities (video_*.mp4), the video_split.py script is expected to produce the following outputs:

1
2
3
4
5
6
7
8
9

mpquic-sbd-vagrant/Workspace/video
├── 2s                    # folder for 2-second segments
|   └── output_dash.mpd   # MPD manifest file describing all 2-second video segments and qualities
|   └── output_dash2.mpd  # MPD manifest file adapted for the SARA ABR algorithm
|   └── segment_*.m4s     # A number of video files containing 2-seconds segments
├── 4s                    # folder for 4-second segments
|   └── ...               # MPD files and segments as in the 2s folder
├── 10s                   # folder for 10-second segments
|   └── ...               # MPD files and segments as in the 2s folder

Warning: do not change the file names, as it causes the video streaming application to fail.

Step 4: Check the location of the video files

The ~/mpquic-sbd-vagrant/Workspace folder is shared between your host and the Vagrant VM. Ensure that the MPQUIC-SBD experimental setup and the video files are placed in this folder. Inside your ~/mpquic-sbd-vagrant/Workspace, you should see two folders:

1
2
3
4

mpquic-sbd-vagrant
├── Workspace/
|   ├── mpquic-sbd/   # source files of the experimental setup
|   ├── video/        # the folder storing the video files

Updated: