Original research objectives
VIZARD will develop a new generation video publishing tool. The main objective of VIZARD is to change the way users deal with video content. A video editor should not limit its creativity to simple time-line representations, but will have the possibility to manipulate the inner structure of a video document (video book) and work with hyper-video representations to build distributed and personalized collections. Publishing personalised video content should be as easy as writing a text document. VIZARD will set a strong focus on usability and mediate the market from the pro-sumer level by the distribution of a light edition as shareware product.
The features of the VIZARD software product are at a glance:
(a) an intuitive desktop to navigate through and preview video footage;
(b) video editing functionality based on the video book paradigm including planning and drafting (pre-production), content selection (structuring & reviewing), and arrangement (comparable to traditional editing);
(c) a video annotation wizard which is based on the video lens paradigm.
The development of compact representations and a temporal segmentation of a video will allow to explore the content and structure of a video in a very fast way. The capabilities of the video editing functionality might be extended using a plug-in approach. The interface will be open and well documented to allow third party suppliers to extend the editing functionality. The performance of the system will be optimised for "faster than real time" processing of video content with special algorithms for editing in the compressed domain. VIZARD will consider standardized formats like XML, MPEG-4 and -7 as the basis of the video format, style descriptions and the data exchange format for all metadata of the annotation wizard. VIZARD will also investigate how to integrate a mechanism to protect the Intellectual Property Rights (IPR) of authors and publishers of video content.
(I) Overall specification - technical description of the system architecture; video style concept, user requirements;
(IIa) Core libraries;
(IIb) VIZARD Prototype;
(III) VIZARD beta version - software with navigation and annotation but limited editing functionality;
(IV) VIZARD version 1.0 - stable tool with the full functionality necessary for video annotation and publishing;
(V) VIZARD version 1.1 - improved and optimised software in terms of performance and price.
VIZARD introduces a new intuitive concept to deal with video content. The software supports users to create a sketch or a story they want to tell. Rather than be constrained by having to follow simple timeline representations, editors are able to manipulate the inner structure of a digital video. This is realized by visualizing a video like a book in sentences, paragraphs and chapters. VIZARD consists of three main components: the VExplorer, the VPublisher, and the VAnnotator. Furthermore a special library supports basic editing operations in the compressed domain for the formats MPEG-2. VExplorer The VExplorer provides the user with a comprehensive "video collection search, organization and management" tool for organizing local content and in order to pre structure material in a repository. The most important feature of the VExplorer is to provide the user with logical views on the local archive on any kind of data (e.g. video, audio, text, images, programs, etc.). These views can be customized as hierarchically structure-able collections representing temporal orders. Next to this the VExplorer enables the user to draft a story or a sketch by selecting and structuring material. During this phase the user can take advantage of the powerful search mechanisms provided by the VExplorer.
Searching for content can be either done on similarity (query by image) or by searching annotations created with the VAnnotator. The entire repository containing the stories can be persistently stored in XML format. Single items or entire sub-trees can be directly imported into the VPublisher in order to be further refined and manipulated. VPublisher The VPublisher is a new-generation storyboard, video editing and video publishing tool. With the VPublisher, the user has the possibility to manipulate the inner structure of a video document we call Video Book, with the result that the publishing of personalized video content is as easy as writing a text document. The basic idea of the VPublisher is to move away from a simple time line representation as it is currently used in commercially available video editing systems. We provide an overall visualization of the object structure by generic layout elements. VAnnotator The VAnnotator provides a flexible and intuitive way for annotating video material. This tool uses the concept of video lenses for providing different views on the materials being annotated. There are lenses for viewing and adding textual annotations and lenses that allow viewing the result of image processing algorithms such as cut detection. The tool follows a timeline paradigm, where each track can have several annotations. All the annotation data is based on the MPEG-7 standard and the output format is also compatible with MPEG-7.
Funding SchemeCSC - Cost-sharing contracts