This section has been divided into 2 sub-sections: "Research" and "Logistics". The former category deals with extensions that relate to ideas about other visual implementation ideas leading to slightly different ways of representing the available data. The latter category addresses implementation issues that would aid in making the application more accessible, more intuitive, and more general to use.

Research

SceneLocation Graph

The ActorSceneLocationPanel was initially designed to hold several different data representations in a tabbed pane. This version of VideoFace ended up with only one such representation, namely the SceneLocation table. An interesting addition to this Panel would be a SceneLocation graph (as opposed to a table). While the table isolates actor information, the SceneLocation graph would combine all the information in one graph. Similar to the SceneLocation table, locations are marked in a column on the left side, and scenes are marked on a row near the top of the panel. Actors are now depicted again as blobs, appearing in all scene/location cells where the actors appear in the episode. Actor blobs of the same actor are now connected with lines throughout the panel. We end up with a graph, from which we may be able to trace an actor's appearance. While the SceneLocation table focused on actor interactions, the SceneLocation graph concentrates on an individual actor's appearance in respect to scene and location.

Animated Graph Evolution

While 2-dimensional graphs and collection of graphs help in understanding some aspect of a film, the time dimension is essentially lost, and only quasi-modeled by means of separating scenes. Since we already have an ordered list of shots from a film, it may be visually interesting to observe the evolution of such a graph by preserving the time element. There are several design choices to be made for this feature. The simpler and computationally light version of this animated graph is to fix the nodes of the graph and fade edges in and out as time elapses or work with a color model to focus on evolving links. The more complicated version is to observe the effect of a developing relationship between actors as time elapses. In this case, nodes would not be fixed, as they are multidimensionally scaled for each new shot. In either case, since the ShotSpreadSheet provides us with time information in terms of frames elapsed, we are able to simulate even the right timing between shots.

Inclusion of Scene Footage

Even though we are mainly concerned with building meaningful graphs for an episode, we should consider including actual scene footage as information that coincides with the graph representations. This could either include an animation-type sequence of scene footage per scene and/or location, or certain key frames can be included as graph nodes in a scene-oriented graph.

Logistics

Generic File Parser

Currently, VideoFace opens data files of not only one specific type (CSV), but also one very particular layout (actor names in column 7, location information in column 11, etc.) A useful change to the file parser would be to implement a way to read virtually any structured text file that contains a certain set of information, i.e. actors, locations, scenes. One possibility is to present the user with a sample section from the to-be-opened file, in which he is to outline what is considered actor, location, and scene information.

Graph Modeler

Eventually this tool should not only serve as a visualizer, but also as a modeler of Graph structures, so as to come up with a data structure of a modeled graph. These data structures can then be compared to other such structures with the final goal of finding similarities. Eventually, this approach should lead to a graph-based search.

Tools and Options

Several tools and options can be added to the existing set. Even though icons exist for Redraw and Display, neither of these tools have been implemented. Similarly, the tools for changing a node's name and an edge's weight have also not been implemented entirely, and have no use at this point.