03.04.2020 - New version available

The Docker container has been extended. A newer version, including new features such as CBR encoding, can be found here: https://github.com/fg-inet/docker-video-encoding

docker-video-encoding

Docker container for DASH video encoding using FFMPEG. The following parameters can be specified:

• crf: value to specify target output quality
• mindur: specify the minimum duration of a segment in seconds (relevant in case of variable segment duration encoding)
• maxdur: specify the maximum duration of a segment in seconds (relevant in case of variable segment duration encoding)
• seglen: specify the fix duration in seconds of all video segments (in case of variable set seglen="var")
• encoder: specify the encoder. supported: x264. planned to be supported: x265, vp9

If the source video shall be splitted into segments of fixed duration, set maxdur=0 and mindur=0; If the source video shall be splitted into segments of variable duration, please set seglen="var".

QUICKSTART (encoding only)

First download an example video:

wget https://service.inet.tu-berlin.de/owncloud/index.php/s/8NA7IFNKN9TgXVA/download -O samples/videos/bigbuckbunny480p24x30s.y4m

Encode the example with fixed segment durations:

sudo docker run --rm -v "$PWD"/samples/videos/:/videos \
                     -v "$PWD"/samples/results/:/results \
                     fginet/docker-video-encoding:latest bigbuckbunny480p24x30s.y4m 41 0 0 4 x264

The command line arguments are:

<video_id> <crf_value> <key_int_min> <key_int_max> <target_seg_length> <codec>

You can find the results in the samples/results folder.

QUICKSTART (with worker and job queue)

Important: Make sure that docker is configured to work without root access.

First make sure you downloaded the video as described about (bigbuckbunny480p24x30s.y4m). Afterwards make a copy of the job template:

cp samples/jobs/00_waiting/bigbuckbunny480p24x30s_job001.txt.tmpl samples/jobs/00_waiting/bigbuckbunny480p24x30s_job001.txt

Start the worker:

python3 worker.py --dry-run --one-job

The following options are used:

• --dry-run: does not call the container, but prints the docker command.
• --one-job: quit after processing one job.

run_workers.sh Script

Use the run_workers.sh script to start a worker on each available core:

cp run_workers_template.sh run_workers.sh

Adapt the paths and parameters in the script and remove the echo from echo screen in the script to activate the workers when you execute the script.

You can stop all running workers after the current job with creating a STOP_WORKERS file in the working directory:

touch STOP_WORKERS

Local Testing

First build the image:

sudo docker build -t fginet/docker-video-encoding:latest .

You can start and enter the build image with:

sudo docker run --rm -it --entrypoint=bash fginet/docker-video-encoding:latest

If you want to test encoding, also add the videos/ and results/ volumes:

sudo docker run --rm -v "$PWD"/samples/videos/:/videos \
                     -v "$PWD"/samples/results/:/results \
                     -it --entrypoint=bash fginet/docker-video-encoding:latest