About
in.fluent is an interactive and customizable actionsport video. it lets you choose what you want to see and creates a never ending video.
We cutted videofootage into really short videos that just show one action e.g. a trick, described what’s happening in the video and stored it in a database. The in.fluent player then acts as a virtual director. It searches for all videos that fits your personal video settings and puts them together to present you your own personal actionsports video.
in.fluent is a project of Daniel
Scheibel and Benjamin Stephan,
LAFKON Publishing GbR.
ABOUT
Influent is a dynamic, interactive online video system. It engages users by allowing them to select their own parameters for custom-made videos and to view only the content that interests them. Detailed descriptions of each stand-alone video clip (that just show one action e.g. a trick) are stored in a database. These descriptions make it possible to categorize the clips into nested genres based on similarities and shared attributes. Scrolling through the database’s categories a user can select their desired video-clip attributes. Users can also determine the key focus areas of the videos by enumerating the percentage of the total video composed from each desired element at each nested level. Utilizing the user’s preferences, video clips are selected, composed, and stitched together, generating a full-length cohesive video. The thematic structure and timing of the resulting video is not random. Rather, the resulting visualizations are based on an analysis of the user’s selected content genre. The video that is generated is continuous and has no pre-determined length. At any time during the video, users can adjust their preferences and interests, thereby modifying the onscreen visualizations. Additionally, there is an integrated music player that allows users to select the accompanying music. If the user does not make any selections, the system will automatically set all necessary parameters until the user begins to indicate preferences.
BACKGROUND
Influent was developed as a collaborative project between Benjamin Stephan and Daniel Scheibel. The first prototype was
presented as a final project in multimedia for the Department of Computer
Science at the University of Applied Sciences Augsburg in April 2007.
This scientific paper discusses a general, technical concept of an interactive
and customizable video system, using sports video as an implementation example.
Further possible areas of application might include news footage, documentaries,
and dynamic motion graphic portfolios.
In July 2007, in cooperation with Red Bull Germany, we were able to continue development of the project. The first influent system launched in June 2008. The system is now publicly available.
We have continued work on the influent system in order to further explore the possibilities of dynamic video, such as how music can be used to dynamically influence visual effects, as well as the integration of relevant annotations and information about the content such as product information.
TECHNIQUE:
Influent consists of three main components:
1. DB/CMS: In order to organize the video content, we created a database structure specifically for Influent that allows you to tag the videoclips with words as well as with complex objects. A recursive content management system (CMS) is used to ensure ease and flexibility when adding new meta-data. The CMS is still in beta, and not yet available to the public.
2. CORE (Virtual Director): The heart of the influent software is a rule-based virtual director that translates the user settings into a continuous and dynamic video stream.
3. PLAYER: The user interface, also called influent-player, is independent from the core-system. According to requirements an individual user interface can easily be created that contains more or fewer of the system’s available functions. Red Bull, for example, requested a small embeddable player that could be integrated into multiple websites (influent.redbull.de, http://www.facebook.com/home.php#…).
The CMS, the core software, and all implemented players are created using Adobe Flex/Flash (Actionscript 3).The communication between client and server uses AMFPHP.
SIMPLE EXAMPLE
Nested Level 1: You choose 60% Sport->Snowboarding and 40% Sport->Freestyle Motocross. Nested Level 2: Within Sport->Snowboarding you can select riders, for example, 50% Rider->Shaun White. With these selections set, the system then starts generating the video timeline and prepares 10 Videoclips. In this example, 6 of them would show snowboarding, out of which, 3 of these specifically show Shauwn White and 4 show motocross. If you do not make any changes the system continues generating clips. The longer you watch the more clips of the same category get pulled out of the video database.
