A Multi-Scale Timeline Slider for Stream Visualization and Control Heather A. Richter, Jason A. Brotherton, Gregory D. Abowd, Khai N. Truong College of Computing & GVU Center Georgia Institute of Technology Atlanta, GA 30332-0280 USA {hrichter, brothert, abowd, khai}@cc.gatech.edu ABSTRACT We present a new user interface technique for the visualization and playback of long media streams decorated with significant events. Our Multi-Scale Timeline Slider allows users to precisely focus on a specific location in a very long media stream or set of streams based on significant events while also retaining the stream’s entire context. KEYWORDS: Timeline slider control, multimedia streams, visualization, focus + context. INTRODUCTION Applications in automated capture and access for ubiquitous computing generate large amounts of captured streams of information (such as video) and events related to these streams (such as when a particular person in the video is talking). Users often browse the media streams using significant events to help pinpoint an exact location to playback. For example, a user might want to review a series of captured meetings that span a period of 3 months in order to discover what “deadlines” were discussed. This may be accomplished by annotating the video timeline with events marking where the word “deadline” was spoken. Thus, the user requires visualization of such events and control over playback of the stream. Our solution entails a timeline slider that scales time consistently, supports focus + context [2], and allows control over stream playback. Traditional approaches to controlling media streams usually involve a simple slider. Moving the slider results in moving to a corresponding part in the media. However, this approach doesn’t scale well to large amounts of media. One pixel in the slider might correspond to several minutes of media. For streams that are hundreds of hours, one pixel can represent several hours of media! We need ways to browse large amounts of media, such as video, yet still have fine control of specific media segments. Infinite focus (and more precise control) can be achieved by using two timelines —one showing the entire media stream, and a second showing a more detailed user- defined subset of the stream. Examples of this are common in video and audio editing programs such as Adobe Premier® and Cool Edit Pro®. In these programs, the only context preserved is the spatial location of the focus region with respect to the entire media stream. This approach does not scale well to timelines that cover large time periods as clusters of events cannot be distinguished and browsed at different levels of granularity. Other research has explored more novel sliders and timelines. The AlphaSlider [1] is a technique for navigating a large number of ordered objects. However, this technique does not show temporal relationships between objects. TimeSlider [3] provides browsing of a large timeline of events while retaining some context. The slider was built using non-linear context —the time scale in the center is fine and linear but is coarse and exponential on the edges. A non-linear representation of time confuses the context of events by distorting temporal relationships between events. Essentially, the non-linear time gives the illusion that everything in the past (or future) is significant by clustering all of the events into one region. MULTI-SCALE TIMELINE SLIDER Our solution, shown in Figure 1, is a multi-scale timeline slider that allows a user to navigate a stream of information and events as well as control the playback of that stream. User controlled, multi-scale zoom is supported through a set of individual timelines that focus the stream context. The user interacts with the timelines by creating focus regions, manipulating existing focus regions, and manipulating the playback control. Multi-Scale Timelines The visualization consists of a series of individual timelines, where each subsequent timeline is a focused region of the previous timeline. Each timeline represents a series of ordered streams. The recessed gray rectangles on the timeline, shown in Figure 1, represent a stream. Each Submitted for review to UIST ’99 To see our widget in action, see: http://fce.cc.gatech.edu/uist99/mts