The way in which OpenVIP core processes multimedia files is based on a so-called network concept (as opposed to timeline concept employed in the GUI). We will use the following diagram to explain what networks are:

This is a scheme of a very simple network, which contains instructions for OpenVIP to load a video file called input.mpg, resize all its frames to 320x180 and save it to output.mpg. Each rectangle in the scheme is called a module. A module usually has some input or output connectors, which link it with other modules.

Let's explain it in terms of our example: There is a module called loader, which is of type Input. This module is indeed responsible of loading the file input.mpg (this is the value of parameter filename). The output connector video0 of loader is linked to the input connector video0 of a module called resize, which is of type VideoFilter. This module actually performs the resize operation - there is a parameter videofilter set to Resize, which tells the module what to do, and a couple of parameters width and height, which specify the new dimensions. Finally, the output connector video0 of module resize is linked to the input connector video0 of another module called saver, which is of type Output and as you expect, it saves the video in the file output.mpg.

Note that the output connector audio0 of module loader is linked directly to the input connector audio0 of module saver, which means that there is no audio processing.

To sum up: Each network consists of modules, which correspond to some actions, and connectors, which link the modules and specify the order of these actions. People who are familiar with graph theory would say that network is an acyclic oriented graph.

Networks are described using XML files which conform to the DTD file openvip-network.dtd. To see that the structure of such a file is really simple let's return to the example from previous section and have a look at its XML version:

<?xml version="1.0" ?>
<!DOCTYPE network PUBLIC
          "-//OPENVIP//DTD Network Format V1.0//EN" 
          "http://openvip.sourceforge.net/dtds/openvip-network.dtd">

<network xmlns="http://openvip.sourceforge.net/network-format" version="1.0">

    <module id="loader" class="Input">
        <param name="filename">input.mpg</param>
    </module>

    <module id="resize" class="VideoFilter">
        <param name="videofilter">Resize</param>
        <param name="width">320</param>
        <param name="height">180</param>
    </module>

    <module id="saver" class="Output">
        <param name="filename">output.mpg</param>
        <param name="format">FFMpeg</param>
    </module>

    <connect module_in="loader" conn_in="audio0" module_out="saver" conn_out="audio0"/>
    <connect module_in="loader" conn_in="video0" module_out="resize" conn_out="video0"/>
    <connect module_in="resize" conn_in="video0" module_out="saver" conn_out="video0"/>

</network>

The interesting part is that between <network> and </network> tags.

Each module is described using a <module> element. The attribute id gives a unique module name (may be chosen arbitrarily), while class describes the type or specialization of the module. Every module can be passed one or more parameters specified using the <param> elements.

A link between the connectors of two modules is introduced using the <connect> element - it simply means that the output connector conn_in of module module_in should be linked to the input connector conn_out of module module_out. Most modules use standardized connector names such as audio0, audio1, video0, video1 etc.

With a network description at hand, let's see how to process it by OpenVIP. This is easily done with the OpenVIP commmand line interpreter. Assuming that our XML network description is stored in file test.openvip, the following command will do the job:

cli test.openvip

If everything goes well, you should see a progress indicator and after the processing finished, you will find the output file in the path specified as a parameter to Output module.

Our example showed how to apply a resize video filter. What other modules are available in the OpenVIP distribution?

The documentation includes a list of all video filters including a description of their parameters. You can of course apply more than one video filter on a single input - the output connector of the first filter is then linked to the input connector of the second filter etc.

Audio filters are easy to use, too. The module class is now AudioFilter and you use audio connectors (audio0 etc.) instead of video connectors. A chain of audio filters is built in the same way as with video filters.

Video transitions and audio transitions belong to class VideoTransition and AudioTransition, respectively. They always have two input connectors and a single output connector.

Input and output plugins are responsible of loading and saving multimedia files. Input modules are of class Input and they have no input connectors. OpenVIP implements several input modules each of which can read different file formats. You usually don't have to specify the input plugin type as OpenVIP tries to autodetect which one to use. It is however possible to deactivate the autodetection and specify which plugin to use in the format parameter:

<module id="loader" class="Input">
   <param name="filename">input.mpg</param>
   <param name="format">FFMpeg</param>
</module>

Output modules are of class Output and they have no output connectors. You must always choose an output plugin using the format parameter:

<module id="saver" class="Output">
   <param name="filename">output.mpg</param>
   <param name="format">FFMpeg</param>
</module>

There are also other useful modules included in the OpenVIP distribution.