The Computer Animated Film: A Dynamic Cartography

Harold Moellering

1973

ACM

ABSTRACT

In the mid 1960's it became evident that the computer animated film offered great potential for developments in the field of cartography. The following example is a realization of that perception. Traffic crash coordinate data from Washtenaw County Michigan for the years 1968-1970 has been assembled and subsequently displayed in a computer generated film animation. The symbol display algorithm is written so that as a crash increases in severity of injury, the larger the symbol and the longer it is displayed. Two crash sequences are displayed: the first in chronological order from 1968-1970 by 24 hour increments, and the second for a composite week by 15 minute increments. The result is a computer animated film about 7 minutes duration.

INTRODUCTION

Several years ago Cornwell and Robinson discussed the potential in cartography of computer animated films (CAF) (Cornwell and Robinson, 1966). Their main point was that computer technology and related cathode ray tube (CRT) technology is increasing and therefore possible contributions to the field of cartography are also increasing. At that time little had been done to apply this technology to cartographic problems relating to CAF production.

This paper presents the methods employed in creating a CAF displaying traffic crashes in Washtenaw County, Michigan for the years 1968-1970. The film itself will then be shown.

Although animated cartography existed many years before practical computers, tremendous hand effort was involved in producing these animations. The procedure of cartographic map animation by hand has been described in the geographic literature (Thrower, 1961). However the ability to portray a temporal sequence in addition to the two (or three) spatial dimensions is a tremendous extention of the ability to map spatial patterns. The heuristic and pedagogic advantages of these dynamic presentations are great and therefore justified the input levels required to produce the handmade film. Today computer and CRT technology has advanced to the point where it is usually much easier to produce a dynamic display by computer than by hand. (Parslow et. al. 1969). This is especially true if the initial data base is in machine readable form at the outset.

A REVIEW OF CAF DEVELOPMENTS

Some of the early developments in CAF displays have occurred at Bell Telephone Laboratories by K. Knowlton and his associates (Knowlton, 1965). They produced films portraying such things as the orientation of a satellite in orbit around the earth, display of certain physical principles as well as a film describing the production of a CAF. Knowlton developed a language called BEFLIX (i.e., Bell Flicks) for use with a Stromberg Carlson 4020 microfilm recorder (Knowlton 1964).

As of this writing films have been produced portraying subjects as diverse as temperature variations in a circuit breaker nozzle and dissolved oxygen variations in a river and others (Phillips, 1972, Sutherland, 1970, Anderson, 1972). CAF displays in geology have been produced of the erosion of the Grand Canyon (Pollack, 1969). Artists have also been interested in the potential offered by the CAF (Csuri, 1970).

GEOGRAPHIC APPLICATIONS OF CAF

Recently Levy, et.al. produced a display of seismicity of the earth during the years 1961-1967 (Levy et.al. 1970). The earth was divided into three geographic regions in which the data was displayed separately. The magnitude of the seismic events was reflected in the size and time duration of the symbols displayed.

In an interesting CAF Tobler presents the growth of the Detroit area for the years 1910 to 2000 (Tobler, 1970). He devised a set of partial differential equations to describe the urban growth and used several decades of past population data to calculate spread coefficients for the time periods. The population distribution was then dispalyed at time intervals of 0.5 and 0.05 years. The genius of this CAF lies in the combining of theoretical formulations with the action of the display in a geographical format, a most impressive accomplishment.

Both of these CAF displays give the observer a simulated real time view of the spatiotemporal phenomena being displayed. By studying these displays in what might be called "a collapsed real time sequence" the viewer develops a feel for the dynamics involved. It seems that such displays as these are a most direct and effective approach in promoting insight to the spatiotemporal dymanics. The following is a further example of such a geographical CAF display.

TRAFFIC CRASHES IN WASHTENAW COUNTY, MICH.

Since 1965 the Highway Safety Research Institute (HSRI) of the University of Michigan has collected data of traffic crashes in Washtenaw County, Michigan. This information has been built into machine readable files which enable the research person to conduct various types of statistical analysis on the data base. However the locational information contained in the file was limited to codes pertaining to the streets forming the intersection nearest to the crash site. Recently Moellering has devised a method of converting these street code pairs into plane coordinates which reflect the location of the crash in Washtenaw County (Moellering, 1972a, 1972b). The system, known as Automap, converts the two street codes of the referenced intersection to the coordinate values of that intersection by a computer based dictionary lookup procedure. That x,y coordinate is then modified to reflect the distance and direction the crash occurred from the referenced intersection if the crash did not occur at the intersection itself. Each reported traffic crash occurring in Washtenaw County from 1968-1970 has been given this geographical processing. (Updating for 1971 and 1972 is now in progress). Currently this Washtenaw crash file with coordinate locations contains over 20,000 cases with about 9 percent of the crash cases having missing data for the crash location. The Automap system also can produce plotter-drawn maps of any subset of the crash coordinate file.

The crash coordinate file also contains 150 other variables collected and generated from the crash reports. These variables contain a great deal of information pertaining to the nature and circumstances of a crash in addition to details of injury patterns of those involved and vehicle damage information. Since the file contains several time variables in addition to the location variables, it is conceptually straightforward to create a CAF to display the data in a spatiotemporal setting. Figure 1 is a simulated frame of the film showing the background map of Washtenaw County, Michigan and crash symbols drawn by plotter.

Any display of this crash data would encompass a constant map background while the crash symbols were displayed with suitable time increments in an appropriate sequence. The entire sequence can then be recorded on film, edited, and prepared for motion picture display.

GENERATING THE FILM

The physical generating of the film was done with the University of Michigan Computer system. The system consists of two IBM 360/67 central processing units operating in tandem plus many peripheral devices. The computer hardware is run under the control of the Michigan Terminal System (MTS), a software processing system. MTS is a large system containing more than 2 million bytes of core storage.

One of the many types of terminal devices which can operate under MTS is a Computek 400/20 storage tube terminal. The Computek CRT is linked to MTS by a voice grade phone line operating at 1200 Baud (about 120 characters per second) output. (For a more detailed description of the time shared operation of the Computek terminal see Phillips et.al., 1970, 1972). The Computek CRT term1nal has a tube face of 1024 by 1024 addressable grid points. The subroutines available can draw any type of figure by addressing this 1024 by 1024 array of points, while commonly used characters and symbols are available as separate subroutine calls (Conklin, et. al. 1971). A program, usually Fortran, can then be written containing these subroutines which generates the desired sequence of frames.

The first step of generating the physical film is running a computer program to create a video display tape, Figure 2. The program reads the Washtenaw data tape containing coordinates which has been sorted into the desired temporal sequence. With appropriate error checking the program reads the data records for each time increment and then generates the necessary computer code for the drawing of the frame. That code is written on a digital video display tape because displaying and filming the frame images on line would be potentially wasteful of computer time. While generating this display tape the program also prints out summary statistics about each time increment generated. This printout is then examined in detail to detect any processing errors.

At a later time the digital display tape is read back through MTS to the Computek CRT terminal. Generally this operation is done late at night to avoid transmission errors over the phone line between the central computer and the Computek. The actual film frames are exposed on high contrast movie film by an Arriflex 16 mm camera operating under computer control. The shutter operating commands are transmitted by MTS from the digital display tape along with the information from which the display film is generated. Essentially the reading of the display tape generates the images on the Computek screen and produces syncronized signals for actuating the camera shutter.

SYMBOL DISPLAY ALGORITHM

The algorithm to display the crash symbols was patterned after that of Pollack (Levy et.al., 1970). Traffic crash reports files in the State of Michigan record injuries as belonging in one of five classes:

• K=Killed
• A=Incapacitating injury
• B=Other visible injury
• C=Complaint of injury
• O=No injury

For convenience C and O injury classes were combined. When reading the data record of any particular crash, the program looked at all injuries and found the worst injury in the crash. The symbol displayed was based on the worst injury as shown in Figure 3. As the seriousness of the worst injury of a crash case increased, the symbol displayed became larger in size as well as being displayed longer in time. There are four symbol sizes ranging from 1/8 inch (on the Computek tube face) to 1/2 inch. Normally the Computek symbols are stick symbols being hollow in the middle. Solid symbols were achieved by drawing the correct symbol size for an injury class plus all of the lower order symbols nested within it. The symbol used for display was a 5 pointed star. As in Figure 3 each time increment (t0...t6...tn) image was exposed on more than one film frame so that individual low order symbols could be seen. For example in Figure 3 a crash level C or O was exposed on three film frames and would be visible for 1/6 second at silent film speed.

A symbol for a class K injury would be displayed for 4 time increments (12 frames) while at each time increment decreasing in size until at the end of the 12 frames it finally disappeared. The star would be ½ inch size for frames 1-3, ⅜ inch for frames d4-6, ¼ inch for frames 7-9, and ⅛ inch for frames 10--12. Other symbols classes were scaled down in size and time duration proportionately. By this method symbols may be lagged in time and size to provide a pronounced visual effect to the viewer.

FILM EDITING AND TECHNICAL PROBLEMS

In the actual filming the symbol sequences and backround map were shot separately, as well as the title sequence. The decision to separate these two integral parts of the display involved the desire to avoid having to transmit the county backround over the phone line for each time increment. As a result display time and the probability of transmission errors was significantly reduced. Later in the film processing the two components of the scene were superimposed to produce a composite sequence. With this method a trivial amount of extra effort is involved in doing the superimposing in color. As a result the map backround is portrayed in a deep blue while the symbols are displayed in a bright yellow. This color combination resulted in the yellow light dominating where the blue and yellow overlapped.

A transmission error can occur whenever a bit of code being transmitted from MTS to the Computek is misinterpreted in some manner. Such an error can result in unwanted 'garbage' being drawn on the tube face. During the actual transmission and filming there is no practical way with the system discussed here to intercept the bad signals prior to being displayed on the tube face and exposed on the film. Instead, the film itself must be edited after the superimposing work is completed. Whether or not a transmission error is actually edited out depends on how serious the error happens to be. As with any data base of this size some cases have missing data codes in essential variables. For example a time increment or coordinate reference could be missing. This missing data is generally the result of two sources: incomplete police reports (law enforcement officers are not professional data collectors) or the inability to interpret information provided for original coding or coordinate building stages of data base construction. Although considerable effort is spent to insure completeness, at the display stage cases with essential variables containing missing data must be ignored.

THE FINAL PRODUCT

The finished film contains two sequences of crash symbols. After the usual title information is shown, a map of Ann Arbor is displayed followed by a map of Washtenaw County. A data reliability diagram is then drawn over the Washtenaw County map to indicate areas of above and below average data reliability. The first sequence displays traffic crashes in the county from 1968-1970 in chronological order by 24 hour increments. In effect it is a "collapsed real time" display depicting these crashes in the same order that they occurred. Each 24 hour increment is recorded on three frames of film. The second sequence depicts the same crash data set by 15 minute increments for a composite week. The sequence begins on Sunday and ends with Sunday again because with a composite sequence such as this the beginning point is arbitrary. Sunday is repeated again for display continuity. Each 15 minute image is recorded on four frames of film. The entire film is approximately seven minutes long.

ACKNOWLEDGEMENTS

The author is indebted to many individuals and organizations at the University of Michigan. The Center for Research on Learning and Teaching provided support in the person of Jon Bauer who wrote the display programs. Professor Richard Phillips of the Gas Dynamics Laboratory provided access to the video display and filming support in the person of John Van Roekel. Professors John Kolars, John Nystuen, and Waldo Tobler of the Department of Geography offered valuable encouragement and advice.

REFERENCES CITED

1. Anderson, Sherwood E., The Scientific Uses of Computer Animation, Proceedings of the National Microfilm Association, Vol. XXI, 1972, pp. 11184-11189.

2. Conklin, James and Barnett, Mark, A Basic Software Package for the Computek Term1nal on MTS, Center for Research on Learning and Teaching, University of Michigan, 1971.

3. Cornwell, Bruce and Robinson, Arthur H., Possibilities for Computer Animated Films in Cartography, Cartographic Journal, Vol. 3 #2, December 1966, pp. 179-82.

4. Csuri, Charles, Computer Animated Film for Art and Education, Proceedings of a Conference on Computers in the Undergraduate Curricula, University of Iowa, Iowa City, June 1970.

5. Knowlton, Kenneth C., A Computer Technique for Producing Animated Movies, Technical Publications, Bell Telephone Systems, Monograph 5112, 1964.

6. Knowlton, Kenneth C., 'Computer-Produced Movies", Science, Vol. 150, November 26, 1965, pp. 116-20.

7. Knowlton, Kenneth C., Levy, Michael A., Pollack, Henry N., and Pomeroy, Paul W., Motion Picture of the Seismicity of the Earth 1961-1967, Bulletin of the Seismological Society of America, Vol. 66, #3, June 1970, pp. 1015-16.

8. Moellering, Harold, Mapping Traffic Crashes in Washtenaw County by Computer, HIT LAB Reports, Highway Safety Research Institute, March 1972, pp. 1-5.

9. The Automatic Mapping of Traffic Crashes, paper 72-CA-lll presented to the ACSM Fall Technical Convention, Columbus, Ohio, October 1972, 22 pp.

10. Parslow, R.D., Prowse, R.W., and Green, R. Elliot (eds.), Computer Graphics: Techniques and Applications, International Computer Graphics Symposium, Brunel University Uxbridge, England, July 1968.

11. Phillips, Richard L., and Geister, Donald E., Production of Computer Animated Films from a Remote Storage Tube 'Terminal, Proceedings of a Conference on Computers in the Undergraduate Curricula, University of Iowa, Iowa City, June 1970.

12. Phillips, Richard L.,and Van Roekle, J.R., Animated Film Production from a Time Shared Graphic Terminal", Proceedings of the International Computing Symposium, CINI Foundation, Venice, April, 1972.

13. Pollack, Henry N., A Numerical Model of Grand Canyon, Four Corners Geological Society Guidebook, 1969, pp. 61-2.

14. Sutherland, Ivan E., Computer Displays, Scientific American, Vol. 222, #6, June 1970, PD. 58-8I.

15. Thrower, Norman J.W., Animated Cartography in the United States Internationales Jahrbuch Fur Kartographie, I, Orell Fussli Verlag, Zurich, 1961, pp. 21-9.

16. Tobler, W.R., A Computer Movie Simulating Urban Growth in the Detroit Region, Economic Geography, Vol. 46, #2, (Supplement), June 1970, pp. 234-40.