Professional Documents
Culture Documents
Storytelling Via Navigation
Storytelling Via Navigation
Storytelling Via Navigation
1 Introduction
Scientific visualizations usually deal with spatial, three-dimensional data sets. An-
imation is a natural way to provide an overview of this type of data. The growing
popularity of video abstracts for summarizing journal publications and conference
proceedings, along with the ubiquity and accessibility of streaming, high-definition
video websites, means that the ability to produce effective, comprehensible video
summaries of scientific work is becoming increasingly important.
However, creating animations using modern scientific visualization software
is often tedious and time-consuming. Animation packages in most current visu-
alization tools use conventional keyframe-based methods, which require training
and practice to use effectively. In addition, certain types of animation sequences,
such as smoothly curved camera paths, are difficult or impossible to create using
keyframes. Thus, scientists who wish to present interesting findings or illustrate
complex phenomena are forced to become proficient with an unfamiliar system
that may not help them achieve their goals.
To help address this problem, we present a method for automatically generat-
ing animations based on user interaction history. As a scientist explores a data
set using a visualization tool they are already familiar with, our system records
all of the user’s interactions, including viewpoint movements, transfer function
modifications, and time spent at each view. This exploration history is then an-
alyzed and used to create an animation that highlights the relevant features of
the data set, without any additional work on the part of the user.
M. Christie and T.-Y. Li (Eds.): SG 2014, LNCS 8698, pp. 1–14, 2014.
c Springer International Publishing Switzerland 2014
2 I. Liao, W.-H. Hsu, and K.-L. Ma
2 Methods
To be able to automate the generation of volume visualization animations, we
should first examine how such animations are manually created. Intrinsically, an
animation for volume visualization is a sequence of images with rendering param-
eters that change over time. A typical way to control these changing parameters
is by using keyframes. A user specifies a number of keyframes with different
configurations at different time points, and the parameters for the frames in-
between are interpolated from the nearest preceding and subsequent keyframes.
This method requires a great deal of user effort to decide which parameter(s) to
keyframe, where to position each keyframe in the animation timeline, whether
to insert pauses between consecutive animated transitions, etc. We present a
new method that allows users to create animations without needing to learn
additional tools (Figure 1).
2.1 Overview
A user exploring a volume data set to find interesting features goes through a
process where they gradually make sense of the data. Similarly, to tell a story
effectively, the order and rate at which pieces of information are revealed must
be carefully paced. Animations of volume data are no exception, and would
likely benefit from using a storytelling approach to present a scientist’s reasoning
process. Thus, the manner in which a user navigates a volume data set as they
are exploring it may serve as a reasonable starting point for considering how
to present the data. As an additional benefit, users would be freed from having
to learn how to use animation tools. Since our target users are domain experts
who are already familiar with and use volume rendering daily, this approach
effectively eliminates the additional training and effort that is currently required
to create animations.
In order to capture a user’s reasoning process in exploring and navigating a
volume data set, we employ the following procedure: 1) A user explores a volume
data set as he/she usually does. 2) During this exploration, the user identifies