[Home Page]

Development and use of an on-line video-analysis tool for physics learning

Jon M Pearce, Michelle K Livett, Susan Rodrigues



There are several computer-based tools world-wide that enable students to carry out analysis of video-clips in the study of Physics. These tools generally allow students to step through the video frame-by-frame, whilst clicking on the location of an object; the data collected are recorded in a table for plotting and analysis.

This paper presents a description of a Web-based video-analysis tool called MotionWorkshop. It is the analysis component of the CAUT funded project Real World Physics. MotionWorkshop, extends the functionality of similar tools by incorporating spreadsheet-style data analysis and numerical modelling into an innovative interface. It is written in the Java programming language so that it can run on any computer across campus, and even at home by those with adequate Internet access. Currently development is taking place to enable QuickTime for Java to be used to present the movies in this analysis tool.

A pilot research project investigated the use of this software in terms of understanding learning strategies used with, and outcomes obtained from, on-line material of this nature. The preliminary results of this research will be presented, together with a discussion of the learning environment, the software, and the challenges in producing a resource of this nature.


A. Introduction

In 1996 a project called Real-World Physics was funded by the Committee for the Advancement of University teaching (CAUT) to develop on-line resources to aid the teaching of physics to first year undergraduate students (Pearce and Livett, 1997). Our philosophy was to give students access to "real world" events through the use of video-clips of the motion of common events (juggler, sprinter, etc.). Part of these resources was a Java applet, MotionWorkshop, which enabled students to analyse the motion in video-clips on a frame-by-frame basis. This paper focuses on the design and construction of this applet, as well as outlining some research being carried out to help us understand the effectiveness of such multimedia-based resources in supporting student learning.

B. Development of a video-analysis tool

During the past few years a number of software programs for the analysis of video-clips have been developed for use in the teaching of physics (see for example the list of Web addresses to projects in the Reference section at the end of this paper). Each varies in the features it offers, but none is Web-based, and none offers many features beyond the basics of entering position-time data from the video, defining coordinate systems, and displaying data in tabular or graphical formats.

For the usual reasons of portability and ease of management, we wanted software that could run over the Web and also offer features well beyond simple analysis. We required a more complex environment in which students could construct numerical (spreadsheet) models to test their understanding of the motions they were analysing by video. Two particularly innovative features required in our original design of MotionWorkshop were:

3D-Spreadsheet: we required an innovative spreadsheet that could record 3-D vector quantities (ie. quantities represented by three numbers: x, y and z), do calculations with them and display them in a 3-D graphing area. This spreadsheet was to have special features to facilitate the process of exploring a numerical model.

Vector display: quantities such as position, velocity, acceleration, force, etc. were to be represented as vectors drawn superimposed on the video, or in a separate display area.

The development of this applet has been slower than originally anticipated due to the vagaries and newness of the Java language (this is news to no-one who has programmed in Java!). Indeed, the project design was conceived in early 1995, before Java was known to the world. Progress is still continuing through 1998 (with support from the Committee for University Teaching and Staff Development as well as from Apple Computer Australia). Below we describe the features that have been implemented to date; vectors are not yet complete nor are the 3-D aspects of the graph display.

Features of MotionWorkshop

The overall aim of MotionWorkshop is to record frame-by-frame position data from a video-clip and then explore the data through spreadsheet and graphical means.

Entering data from the video-clip

Students can load a video-clip, or a still multi-exposure image, into MotionWorkshop from a database of resources on the Web. Their initial task is to record the (x,y) position data for the object(s) they wish to study. This is done by clicking on the object(s), frame by frame, and the data are automatically entered into the spreadsheet. The figure shows the multi-exposure still image of a golfer (courtesy of the Harold and Esther Edgerton Foundation), with the currently selected point labelled. The spreadsheet values are "hot" in that editing them will affect the points displayed on the video, and vice versa.

Students can choose to work in Cartesian coordinates or polar coordinates, as shown in the figure, and swap between the two. They can also choose to display the other kinematic quantities directly derived from these data, namely velocity and acceleration, in Cartesian or polar form, either (x,y) components, or (radius, angle) components.

Displaying graphs

One or two graphs can be chosen to be displayed by clicking on the labels on the graph axes. These graphs can be independently shifted and scaled by simply dragging the axes on the screen. This has proved to be a most intuitive way to manipulate graphs, but was a major headache to the programmer to implement! The figure shows a graph of angular displacement and velocity from the golfer image.

Modelling with the spreadsheet

The spreadsheet is almost fully functional, with a few features still being implemented. Whilst it provides most of the usual spreadsheet functions useful to a science student, it adds to these several new features to facilitate numerical modelling:

Smoothing: since velocity and acceleration are calculated by the spreadsheet from data that might have inaccuracies, there is a need to smooth these data by choosing a variable time interval over which to do the calculations. This choice is made by varying the slider to the left of the graph (see the figure above). This gives students the chance to explore different smoothing settings in a very interactive way.

Time-slicing: most video-clips that students will analyse have a frame rate of 25 frames per second (0.04 seconds per frame). However, if the students also wish to model the motion, they will probably require a "time interval" value much smaller that 0.04 seconds for accurate results. That is, they will need to "slice" the time intervals determined by the video frame rate into smaller intervals. We addressed this problem by introducing a "time-slicing" concept whereby the student can insert extra rows into the spreadsheet between the existing video data rows. The spreadsheet rows can also be collapsed down to display to display every 2nd, 3rd, 4th, or nth row. Hence the extensive number of rows required for accuracy can still result in a spreadsheet that will fit within a small screen.

Changing variables: one aim of numerical modelling might be to compare the output of the model with data recorded from a video. Often variables in the model will initially be unknown and a trial-and-error approach might be required to "tweak" them. To do this the students can attach any spreadsheet variable to a slider and watch the spreadsheet update as they play with the value (albeit rather slowly for a large spreadsheet!).

Development issues

MotionWorkshop is an extremely complex piece of software especially for a Java applet. It is currently performing well, but there are several issues still to be addressed.

As alluded to earlier, working with Java has been a challenge. The language is young, unstable and rapidly developing. Differences exist between different platforms and between different browsers within those platforms. Currently the MotionWorkshop applet runs reliably on Internet Explorer and Netscape 4.03 (with the JDK1.1 patch) on the PC platform, and Internet Explorer 4 on the Macintosh. However, this undesirable situation is improving rapidly and hopefully, very soon, any current version browser will run MotionWorkshop in a consistent manner.

At the time of writing, it is not possible to run QuickTime movies from within a Java applet across these platforms. The "movies" that we have run so far have comprised a sequence of GIF files generated from QuickTime movies. Apple has a very active project group developing "QuickTime for Java" and, as soon as this becomes more stable, we will implement it for the project.

Skilled Java programmers are very hard to find. We have been fortunate in finding an extremely talented one, but, as is often the case, we have to share him with other projects and hence progress has been slower than we originally anticipated. However, developments are still underway and we expect to be able to implement all the planned features in the near future.



Researching the use of video-analysis in learning

Having developed MotionWorkshop, we had the opportunity to interact with students using it in order to explore the learning that takes place through computer-based and laboratory-based interactions. Although this research is not complete, we present here an overview of the ongoing project (funded by the Australian Research Council).

The question

Whilst using such tools as MotionWorkshop is fun, there is still much debate about the effectiveness of video as a medium to support student learning compared with real, hands-on laboratory experiences. (For discussions on using microcomputer-based and video-based resources see Beichner 1990, Escalada & Zollman 1996, Redish et al 1997, Thornton and Sokoloff 1990.)

It is often difficult, or even meaningless, to make such comparisons since the intended explorations and outcomes from a video-based lab session might be very different from a traditional lab. However, we had the opportunity to shed some light on this issue through duplicating a traditional physics lab exercise with one using MotionWorkshop. The traditional lab exercise used motion sensing devices to record the motion of a cart along a benchtop, and a ball bouncing up and down. (A description of this ultrasonic motion sensing technique is given in Pearce, 1993.) The controlling software displayed graphs of position, velocity and acceleration in real time. This microcomputer-based laboratory (MBL) style of lab has been proved to be very effective in promoting learning in a variety of other studies (see, for example, Redish et al 1997).

Two of the issues we were keen to examine were: did the fact that some students were pushing "real" carts promote more robust learning outcomes, due to the kinaesthetic nature of the activity, compared to using a video of the motion? Did the fact that some students were clicking frame-by-frame on the video give them a different understanding of the motion?

Methodology, Issues and Outcomes

For our study we designed a laboratory exercise that was almost identical to the "real" one except that the initial data of the cart’s motion were collected from video, as were the ball's data.

It should be stressed here that this was not regarded as an ideal scenario for the use of computer-based video analysis. The "real" laboratory exercise has all the benefits of hands-on manipulation of equipment and real-time data collection. However, it was an opportunity to observe students’ learning while using this medium, and hopefully understand better the learning strategies they employed.

From the 14 groups of students comprising a first year physics laboratory subject, we selected 3 groups (47 students) to take the alternative video-analysis laboratory, leaving 11 groups (160 students) to take the traditional laboratory exercise. All students were given pre- and post- tests. Twelve students were interviewed one week later.

A careful analysis of test results and interviews is still being carried out. However, our tentative findings reinforce the contention that learning tasks that actively engage students are most likely to achieve change in student thinking. Support for this view comes from students' comments about the learning activities. In the activity in which students analysed the motion of the cart, those who actually pushed the cart and observed the graphs being produced in real time commented on the role of pushing the cart:

"I could relate more to the graph on the computer as a result of my own movement",

"moving objects allowed me to see what different movements would do to the graphs",

"controlling the object helped me to understand how graphs were formed".

Where it was not possible for students to experience kinaesthetically the motion being graphed, for example the bouncing ball, video-analysis could provide a measure of active engagement. When interviewed about the bouncing ball analysis, the action of clicking on the ball’s successive positions provided kinaesthetic reinforcement of the slowing down of the ball as it rose to its highest point, and its speeding up on the way down.

The pre- and post-tests were intended to assist us in determining changes in students’ understanding of graphs of motion as well as the relationship between forces and changes in motion. We did not find significant differences between the role of the video-analysis and motion sensing activities. While students who undertook the video-analysis task performed more successfully on the end of semester assessment, further research focussing on these concepts is required to confirm the indications from this preliminary research.

Overall this leads us to the tentative conclusion that the choice of using a video-analysis tool should focus on motion which is not amenable to active student engagement in the motion.


Work is still ongoing in developing the MotionWorkshop applet. This work largely involves refining the user interface and implementing vectors and QuickTime for Java. The research effort has raised several questions regarding how such a tool can affect learning and the place multimedia generally has in supporting meaningful learning.



R J Beichner, The Effect of Simultaneous Motion Presentation and Graph Generation in a Kinematics Lab, J. Res. Science Teach. 27, 803—815 (1990).

L T Escalada and D Zollman, An Investigation on the Effects of Using Interactive Video in a Physics Classroom on Student Learning and Attitudes, Journal of Research in Science Teaching 34, 476-489 (1996).

J. M. Pearce, Measuring Motion Using the Macintosh Computer, Aust. Science Teachers’ Journal, 39, No. 2, pp 44-51. (June 1993).

J. M. Pearce and M. K. Livett, Real-World Physics: a Java-based Web Environment for the Study of Physics, Proceedings of AusWeb97, Brisbane, (July 1997).

E. F. Redish, J. M. Saul, and R. N. Steinberg , On the Effectiveness of Active-Engagement Microcomputer-Based Laboratories, American Journal of Physics, 65 (45) (Jan 1997).

R K Thornton and D R Sokoloff, Learning Motion Concepts Using Real—Time Microcomputer—Based Laboratory Tools, Am. J. Phys. 58 (9) (1990).


Web addresses for video-analysis software.

R Carlson, World in Motion, http://members.aol.com/raacc/wim.html

R Beichner, VideoGraph, http://www.aip.org/pas/videograph.html

P Laws, VideoPoint, http://www.lsw.com/videopoint

Learning in Motion, Measurement in Motion, http://www.learn.motion.com/lim/mim/mim1

J Pearce & M Livett, MotionWorkshop, http://www.science.unimelb.edu.au/rwp/MWdemo.html

TERC, CamMotion, http://hub.terc.edu/terc/view/view_homepage.html

[Back to J Pearce Home Page]

These pages are maintained by Jon Pearce ( jonmp@unimelb.edu.au), Department of Information Systems. The opinoins on them do not necessarily reflect those of the University of Melbourne. Tel: (613) 8344 1495 Fax: (613) 9349 4596. Last update: September 16, 2003 .