From PiTiViWiKi

You can think of this article as a complement to the more traditional API reference and Design Docs. Rather than laboriously enumerate class relationships and API speficications, we will follow the implementation of two features from the initial concept to the final implementation. Along the way, we introduce various components of PiTiVi's design on an as-needed basis. Hopefully, you'll emerge on the other side with a better understanding of how to modify PiTiVi.

The two features we will be examining are interpolation curves and still images. Interpolation cuves are an example of an intrusive change that expands PiTiVi's fundamental capabilities. We will see how adding capabilities to TrackObjects creates rippling changes in many other parts of PiTiVi. Still Image support is a good example of how to implement new SourceFactory objects and how to implement new GStreamer objects in python.

Contents 1 Interpolators 1.1 Proof Of Concept Demo 1.2 UI Interaction Demo 1.3 Defining the API 1.4 Dummy Back-end Implementation 1.5 Develop and Test the UI 1.6 Making the Back-end Do Something 1.6.1 Providing Access to Required Elements 1.7 Extending the Project Formatters 1.8 Integrating into the Undo System 1.9 Unsolved Issues 2 Still Images 2.1 Proof of Concept Demo 2.2 Initial Implementation 2.2.1 Changes to Discoverer 2.2.2 New SourceFactory Type 2.3 Issues with the Freeze Element 2.4 Writing a Python Freeze Element 2.5 Final Implementation

Interpolators

Many NLEs provide a mechanism for editing time-varying values, such as audio volume and clip opacity. GStreamer exposes this ability through gst.Controller and supporting classes. The design documents for pitivi show a number of mock-ups which include keyframe curves overlaid directly atop track-objects in the time-line. We will develop this feature using the gst.Controller API to provide the actual functionality.

Proof Of Concept Demo

First, we should familiarize ourselves with the gst.Controller API. Let's create an automated of existing videobalance demo.

Exerpt 1

The trick now is to control one or more of the videobalance element's properties with a gst.Controller. First we create a controller, and bind it to the element's properties.

Excerpt 1

Then, we add a series of control points. Since we're trying to test out this functionality, let's choose a test pattern that will be easy to detect.

Image:Interpolator test pattern.png

The code to add this curve is pretty straightforward:

We create the 

UI Interaction Demo

This next demo is a prototype keyframe interface.

Defining the API

Dummy Back-end Implementation

Develop and Test the UI

Making the Back-end Do Something

Providing Access to Required Elements

Extending the Project Formatters

Integrating into the Undo System

Unsolved Issues

Still Images

Proof of Concept Demo

Initial Implementation

Changes to Discoverer

New SourceFactory Type

Issues with the Freeze Element

Writing a Python Freeze Element

Final Implementation