HLS Video on Demand streaming

First of all, let’s explain shortly what HLS is.

HTTP Live Streaming (also known as HLS) is an HTTP-based media streaming communications protocol implemented by Apple. Since it requires only standard HTTP transactions, HTTP Live Streaming is capable of traversing any firewall or proxy server that lets through standard HTTP traffic, unlike UDP-based protocols such as RTP. This also allows content to be delivered over widely available CDNs. In a few words, HLS works by breaking the overall stream into a sequence of small HTTP-based file downloads. At the start of the streaming session, the client downloads an extended M3U (m3u8) playlist containing the metadata for the various sub-streams (called TS files) which are available. You can read more about HLS Architecture on Apple Developer website.

Currently, there are two main solutions for streaming Adobe RTMP streaming, that we covered in the previous blog post and Apple HLS streaming. However, both technologies allow you to play your video as you record it, automatically adjust video quality to available bandwidth, and seek to different parts of a video.
The major differences between the two technologies are, that while Adobe RTMP works only in Flash and requires you to have a dedicated RTMP server installed,  Apple HLS works with both Flash and HTML5 and can be used with an ordinary web server.
Furthermore, when Apple decided to drop Flash support for iOS (the affected devices are iPhones, iPads, laptops, etc) the developers had to think to a solution for the users of these devices.

The first step in our tutorial it is to compile nginx web server with rtmp modules. We described the necessary steps also in our previous post , but we will mention them again here.

cd ~
mkdir nginx
cd nginx
# for compiler and git
apt-get install git gcc make
#for the HTTP rewrite module which requires the PCRE library
apt-get install libpcre3-dev
# for SSL modules
apt-get install libssl-dev
git clone https://github.com/arut/nginx-rtmp-module
wget http://nginx.org/download/nginx-1.4.3.tar.gz
tar zxpvf nginx-1.4.3.tar.gz
cd nginx-1.4.3
./configure --add-module=/root/nginx/nginx-rtmp-module/ --with-http_ssl_module --prefix=/usr/local/nginx-streaming/
make
make install

We have to change the default nginx configuration file:

cd /usr/local/nginx-streaming/conf
mv nginx.conf nginx.conf.bkp
nano nginx.conf
 
worker_processes  1;
 
events {
    worker_connections  1024;
}
 
http {
    access_log /var/log/nginx/access.log;
    error_log /var/log/nginx/error.log;
    include       mime.types;
    default_type  application/octet-stream;
    sendfile        on;
    keepalive_timeout  65;
 
    server {
        # in case we have another web server on port 80
        listen       8080;
 
        location /hls {
          types {
            application/vnd.apple.mpegurl m3u8;
            video/mp2t ts;
          }
          #where the m3u8 and ts files are
          alias /var/www/hls;
        }
 
        location / {
            # here we can put our website
            root   /var/www/html;
            index  index.html index.htm;
        }
 
    }
 
}

We compiled ffmpeg according to this guide.
After we finished to compile ffmpeg, we created an executable also for qt-faststart (we will explain later the use of this tool)

cd ~/ffmpeg_sources/ffmpeg/tools
make qt-faststart
cp qt-faststart $HOME/bin

Now, we have all the necessary tools to start streaming… but, we don’t have the files. We are going to use the well known movie of Big Buck Bunny (right click and save the file in /var/www/hls folder on your webserver).

In order to prepare the ts files (media segments) and m3u8 playlist we are going to use ffmpeg. We compiled according with the guide recommended above, and we have ffmpeg executable in $HOME/bin/ffmpeg. You can define in FFMPEG variable the path to your own ffmpeg executable.

cd /var/www/hls
FFMPEG=$HOME/bin/ffmpeg
#pay attention
FILENAME=bbb
$FFMPEG -i $FILENAME.mp4  -codec copy -map 0 -f segment -vbsf h264_mp4toannexb -flags -global_header -segment_format mpegts -segment_list $FILENAME.m3u8 -segment_time 10 $FILENAME-%03d.ts

Unfortunately, the streaming doesn’t work all the time, and we need to re-order the MP4 “atoms”. These atoms are the meta-information about the movie (like the timescale, duration, other characteristics). For fast delivering over the network these meta information should be at the beginning of the file, not at the end. You can read more about atoms in the following article Understanding the MPEG-4 movie atom.
To move the atoms at the beginning of the file, we will use qt-faststart tool in this way.

FASTSTART=$HOME/bin/qt-faststart
mv bbb.mp4 bbb_original.mp4
input=bbb_original.mp4
$FFMPEG -i $input -vcodec libx264 -vprofile high -preset slow -b:v 500k -maxrate 500k -bufsize 1000k -vf scale=854:480 -threads 0 -acodec aac -ac 2 -strict experimental -b:a 128k bbb_tmp.mp4
$FASTSTART bbb_tmp.mp4 bbb.mp4

and now let’s execute again the previous command:

$FFMPEG -i bbb.mp4  -codec copy -map 0 -f segment -vbsf h264_mp4toannexb -flags -global_header -segment_format mpegts -segment_list bbb.m3u8 -segment_time 10 bbb-%03d.ts

Anyway, sometimes (if the video file is broken or not standard), qt-faststart can fail with some side effects (high CPU load). Fortunately, there is a python script that can handle better the move of the atoms.

We can see the result of our work, using a m3u8 player (eg: VLC) and opening the stream from Media->Open Network Stream and in URL field pasting: http://mydomain.com:8080/hls/bbb.m3u8 (replace mydomain.com with your domain name or IP).

Hit the Play button and… the video will start to be played.

If you want to play the movie in the browser, you can read the post about How to play HLS with JW Player.

You can see an example here – if the server is up , or put in your VLC the URL http://cdn.razvantudorica.com/hls/2/bbb.m3u8

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.