The context of this PhD project is the area of multimedia content management, in particular interaction with home videos. Nowadays, more and more home videos are produced, shared and edited. Home videos are captured by amateur users, mainly to document their lives. People frequently edit home videos, to select and keep the best parts of their visual memories and to add to them a touch of personal creativity. However, most users find the current products for video editing timeconsuming and sometimes too technical and difficult. One reason of the large amount of time required for editing is the slow accessibility caused by the temporal dimension of videos: a video needs to be played back in order to be watched or edited. Another reason of the limitation of current video editing tools is that they are modelled too much on professional video editing systems, including technical details like frame-by-frame browsing. This thesis aims at making home video editing more efficient and easier for the non-technical, amateur user. To accomplish this goal, an approach was taken characterized by two main guidelines. We designed a semi-automatic tool, and we adopted a user-centered approach. To gain insights on user behaviours and needs related to home video editing, we designed an Internet-based survey, which was answered by 180 home video users. The results of the survey revealed the facts that video editing is done frequently and is seen as a very time-consuming activity. We also found that users with low experience with PCs often consider video editing programs too complex. Although nearly all commercial editing tools are designed for a PC, many of our respondents said to be interested in doing video editing on a TV. We created a novel concept, Edit While Watching, designed to be user-friendly. It requires only a TV set and a remote control, instead of a PC. The video that the user inputs to the system is automatically analyzed and structured in small video segments. The editing operations happen on the basis of these video segments: the user is not aware anymore of the single video frames. After the input video has been analyzed and structured, a first edited version is automatically prepared. Successively, Edit While Watching allows the user to modify and enrich the automatically edited video while watching it. When the user is satisfied, the video can be saved to a DVD or to another storage medium. We performed two iterations of system implementation and use testing to refine our concept. After the first iteration, we discovered that two requirements were insufficiently addressed: to have an overview of the video and to precisely control which video content to keep or to discard. The second version of EditWhileWatching was designed to address these points. It allows the user to visualize the video at three levels of detail: the different chapters (or scenes) of the video, the shots inside one chapter, and the timeline representation of a single shot. Also, the second version allows the users to edit the video at different levels of automation. For example, the user can choose an event in the video (e.g. a child playing with a toy) and just ask the system to automatically include more content related to it. Alternatively, if the user wants more control, he or she can precisely select which content to add to the video. We evaluated the second version of our tool by inviting nine users to edit their own home videos with it. The users judged Edit While Watching as an easy to use and fast application. However, some of them missed the possibility of enriching the video with transitions, music, text and pictures. Our test showed that the requirements of overview on the video and control in the selection of the edited material are better addressed than in the first version. Moreover, the participants were able to select which video portions to keep or to discard in a time close to the playback time of the video. The second version of Edit While Watching exploits different levels of automation. In some editing functions the user only gives an indication about editing a clip, and the system automatically decides the start and end points of the part of the video to be cut. However, there are also editing functions in which the user has complete control on the start and end points of a cut. We wanted to investigate how to balance automation and user control to optimize the perceived ease of use, the perceived control, the objective editing efficiency and the mental effort. To this aim, we implemented three types of editing functions, each type representing a different balance between automation and user control. To compare these three levels, we invited 25 users to perform pre-defined tasks with the three function types. The results showed that the type of functions with the highest level of automation performed worse than the two other types, according to both subjective and objective measurements. The other two types of functions were equally liked. However, some users clearly preferred the functions that allowed faster editing while others preferred the functions that gave full control and a more complete overview. In conclusion, on the basis of this research some design guidelines can be offered for building an easy and efficient video editing application. Such application should automatically structure the video, eliminate the detail about single frames, support a scalable video overview, implement a rich set of editing functionalities, and should be preferably TV-based.
|Qualification||Doctor of Philosophy|
|Award date||7 Oct 2009|
|Place of Publication||Eindhoven|
|Publication status||Published - 2009|