Track and field games were rather popular during the eighties. The brutally simplistic joystick waggling of Activision’s Decathlon and the slightly more sophisticated button bashing of Konami’s Track & Field started it all in 1983. While the genre hasn’t exactly been thriving in the nineties and into the new millennium, Konami has continued to release games under their Track & Field brand and every Olympic year they have to face a few competitors.
Most game genres have evolved dramatically since the early eighties. Pole Position may still be a charming and fun game (and even good looking) but in terms of complexity and depth it cannot compare to the racing games of 2002. The implementation of real world (exaggerated or not) physics and mechanics should in part be credited for this. Track and field games, however, have not undergone the same path of evolution. Of course, 3D-visuals have entered the screens, but as the track & field events in most cases are executed along a straight track or runway (i.e. in one dimension), the 3D-visuals have been no igniting spark for game play innovations.
While racing games are often developed by people with a strong interest in motor sports and cars, I strongly doubt that the developers of track and field games have had either a strong interest in the sport of track & field or the biomechanics and technical execution of the events. But I have, and because of that we were able to add several game play innovations to the track and field genre in our game: Fila Decathlon.
Just a few years before the release of Decathlon and Track & Field, I had started training track and field as a young kid. Also a videogame fanatic, I was naturally spellbound by the sight of Konami’s arcade cabinet. The gameplay didn't’t disappoint. It successfully reproduced one fundamental attraction of real life track & field – the lure of trying to beat records whether that would be your personal best or the world record.
And then there was the congenial control method - the faster you waggle the joystick (or press the buttons), the faster you run. The game actually become a test of speed and endurance just like the real sport, but a test that even a couch potato, softened by years of videogaming, could be competitive at.
After having played several track and field games obsessively at home, I of course thought of things that could have been done better or differently. As a track and field athlete myself, I was more sensible to certain shortcomings than others were. Still, most people who had ever watched athletics on television felt that the ridiculous results they were able to achieve in many track and field games, like several seconds below the world record in 100m, took away from the experience. In the same way the very unconvincing animation and technical execution of some of the events were disturbing even if didn't’t necessarily affected the actual gameplay.
However, some gameplay related problems were apparent. In many events you will rather soon reach your maximum potential. Quite simply there are too few factors that influence the end result. The speed factor is of utmost importance. You will probably find that you are able to reach a certain maximum speed and that you are not really able to improve it by training. The running speed is the major factor also in the jump and throw events and as the other factors are few in number and rather low in resolution (limited number of outcomes) the longevity of the game suffers.
So in short, the learning curve is short and this makes for a game fun to pick up and play but too easy to master.
When I first started to think about making a track and field game, I had no real intention to change the rigid formula of how the single events were supposed to work established by previous games. My vision was mainly concerned with the game structure and the game visuals.
I wanted to re-use the decathlon structure of Decathlon as it naturally provides complexity and depth (even if you will often make a more or less perfect jump or throw, you will certainly not be able to do so ten events in a row).
I wanted several CPU controlled competitors to participate in the decathlon. The decathlon should not be only about gaining points; it should also have an element of real competition. Previous track & field games have often included a much praised multiplayer mode, while the single player experience has been seriously lacking due to very poorly implemented CPU controlled competitors. You never see the CPU controlled athlete jump or throw. They participate only in the running events, more or less like visual filling, and as result they remain completely anonymous to the player. In our game I wanted the single player experience to mimic, and in certain aspects even surpass, the multiplayer experience. It should not be possible to distinguish a CPU controlled athlete from a human controlled one. Of course, you should be able to watch the CPU controlled athletes jump and throw, and their action in all events should be the result of mimicking the input from a human player. All the CPU controlled athletes should have a unique profile, e.g. some are good at the javelin, some are good at the hurdles. Two individuals who are equally fast at the 110m hurdles may still be very different. One of them may be the better 100m runner (i.e. faster) but a poor hurdle technician.
I also wanted an interesting competition structure. The ultimate goal should be to win the World Championship. To qualify for the World Championship, however you must place among the first two in the National Championship. There is a set of athletes for each nation and as only the two first qualify for the World Championship, the competition is naturally much tougher at the World Championship compared to the National Championship.
And I wanted realistic result levels. An exceptional player should be able to score about level with the current decathlon world record of 9026 points. The single event world records in track and field would together result in a decathlon score of 12487 points so even an exceptional player should hardly be able to run 100m at the world record time of 9.78s and much less below 8.00s like in many other track & field games.
I wanted realistic animation and visuals that would reproduce the feel of real track and field. As the game should be sprite based, none of the 3D track and field games inspired me and, in addition their animation was not really realistic. A few previous sprite based games have had fluid and elegant animation if still not very realistic. Summer Games and Summer Games II by Epyx are two early examples from the mid-eighties. I decided to opt for a similar style but naturally my intention was to improve the animation considerably.
I wanted to keep the original button mashing control method of earlier track and field games. Some games have, in a misdirected attempt to evolve the genre, abandoned the classical brutal control method in favor for more sophisticated (in a superficial sense) methods. In my opinion the physical quality of the original control method involves the player to a much higher degree and is much closer to the spirit of the sport that it tries to simulate.
The Speed Algorithm
With this vision clear in my mind I started to consider the design of the actual events. As 100m is the first event of the decathlon and as eight of the ten events in the decathlon involve running, it was natural to start with the 100m.
At this time in my life, I had taken up track & field training again after a break of some years. Now older, I had a more analytical interest in the sport. I studied video recordings of competitions and read scientific studies.
With realism in my mind, I began to think about the algorithm for running speed. You cannot simply let a certain button pressing frequency correspond to a certain running speed. The player is able to reach maximum button pressing frequency at once from the shot of the gun. A runner, however, reaches top speed only after 40-50 m of acceleration. We can achieve this easily by letting each button press increase the running speed and the higher the running speed, the smaller the increase in speed. If we also include a constant deceleration factor, a certain button pressing frequency will eventually result in a certain maximum running speed.
By adjusting the parameters of this algorithm we should be able to let a player of a certain skill run the 100m at a certain time. We found that it should be possible to simplify the player input to a constant button pressing frequency from start to the finish. By simple statistical analysis of the decathlon series by the top 20 performers of all time, we found that 10.56s and 11.03s at 100m correspond respectively to 9000p and 8000p in the decathlon. As 9000p roughly equals the world record, the button frequency of a 10.56s race should only be attainable for gifted player. As 8000p would roughly match the score needed to qualify to the World Championship, the corresponding button pressing frequency should be attainable for most players.
To ensure that maximum speed is reached in the right part of the race (i.e. after 40-50m) further balancing of the parameters is necessary. A wacky velocity-distance curve could easily spoil the effect of an otherwise realistically and carefully done animation of the runner. Still, you could argue that the fine tuning of the velocity-distance curve is only a visual matter and thus of minor importance. If we didn’t aim for perfection, I would be inclined to agree. However, we should naturally use the same algorithms and parameters for the running speed in the long jump and as the velocity-distance curve has a more profound influence on the long jump, there are strong reasons to have it properly done from the beginning. The time-velocity curve determines when top speed is reached and consequently it determines the length of an optimal run-up. The velocity-distance curve also determines the top speed, which is the major deciding factor of the length of the actual jump
The animation of the running stride
Let’s then turn to the visual representation of running. It should be pretty simple, shouldn’t it? Just make an animation cycle that constitutes two running strides and make sure to draw enough frames for smooth movements. Well, it will certainly look odd if the runner goes directly from the crouched starting position to an upright running position. Games with a comical graphical style like the original Track & Field can get away with this but it would clash very badly with an otherwise realistic animation. A realistically animated acceleration phase will require an awful lot of frames and plenty of work considering that our intention is to use hand drawn sprites. Fortunately we should be able to use the same animation in the 110m hurdles and the 400m and perhaps also part of it in the long jump (you do not start from starting-blocks in the long jump).
A few words on how the animation was actually made: I guess you could say that we used a rotoscoping technique. I filmed athletes for every animation sequence, extracted the video frames, cut out the frame of the athlete, manipulated size, colors and lights and finally I manually drew/colored clothes, shoes and the head of the athlete along with a lot of other manual modifications. No fancy camera equipment was used. To maintain a proper side view perspective, I filmed from a long distance with a lot of zoom and occasionally I had to take several shots from different viewpoints.
Sliding feet is a very common visual shortcoming in all kinds of videogames. Track and field games are no exception. The quality of the visuals and animations of many sports games today may at first glance be mistaken for real television footage but sliding movements will destroy the illusion and reveal its artificial origin. I am proud to say that there are no sliding movements at all in Fila Decathlon.
How do we avoid sliding feet during the running animation? If every frame change corresponds to a certain forward distance covered, we can simply let the frame changes be distance dependent instead of time dependant. The animation cycle of one running stride then corresponds to the sum of these distances, which implies that the time needed to complete a running stride is inversely proportional to the running speed. That is, stride frequency increases with running speed. In reality, faster running is achieved by an increase in both the stride length and the stride frequency. A running stride can be divided into the ground support phase and the airborne phase. If the frame changes during the ground support are distance dependant and the frame changes during the airborne phase are time dependent, the outcome should be the desired – no sliding feet and a stride length as well as a stride frequency that increase with the running speed. (As an aside: the stride frequency of elite sprinters is about 4.5 - 5 strides/second and at top speed their stride length typically varies between 2.2 – 2.4 meters.)
Ok, as you can see a lot of thought and work were done to optimize the realism of running. But did it have any significant effect on the playability of the 100m event? Not really, what makes the 100m more interesting in Fila Decathlon than in most other games are probably the well balanced result levels and the challenging computer controlled athletes.
A short analysis of the factors that determine the outcome in the 100m would result in the diagram below.
Factor one, the reaction time, is of course present in all track and field games. A neat addition in Fila Decathlon is that the reaction time is displayed on a clock of the in-game graphics and just like in real track and field a reaction time below 0.10s is considered to be beyond the human limit and is penalized by a false start warning. (In the decathlon you are allowed two warnings, after the third warning you are disqualified.)
Factor two, the button pressing frequency, is more complex as it stretches over time. A bit simplified, you could say that it consists of the ability to achieve a certain button pressing frequency and of the ability to keep this button pressing frequency during the whole race. With regard to the playability, it is really not the realism of the speed algorithm but its sensibility to the slightest change of pressing frequency that matters.
Factor three, the timing of the finishing dip over the finish line, is not present in all track and field games. In reality a finishing dip, even if perfectly timed, improves your time very little. We actually exaggerated the effect of the finish dip to add significance to this factor, that is, we departed deliberately away from realism.
The Long Jump
Let’s now turn to the Long Jump, the second event of the decathlon. The long jump is a complex event that consists of a running part (the run-up) and the jump itself.
The gameplay elements of the long jump used in Konami’s original Track & Field have remained intact in the majority of its successors. You build up speed during the run-up. The faster the speed at the time of the jump, the longer you jump. The jump is executed by pressing the jump button. How long the button is held down before it is released determines the angle of the jump. There is an optimum angle and consequently there is an optimum amount of time to hold the button pressed. The timing of the press also matters. If you press the jump button too late and cross the take-off board, it will be a foul. If, on the other hand, you press the button too early, distance is lost as the length of the jump is always measured from the front line of the board.
We can make a diagram of the factors that determine the actual length of the jump.
Surprisingly, the number of factors is the same as in the 100m. However, compared to the 100m, the speed factor is not as dominant in the long jump. Especially the jump angle has a very strong influence on the length of the jump. Also, the speed factor is different from the speed factor in the 100m as it is only the speed at the time of the jump that matters. Still, the limited number of factors makes the event rather stale after any having played it for some time.
Now, what about the realism? Well, actually it is not even close to the real thing. The diagram presented is a very poor representation of the factors of a real long jump. Also, the feel of performing the jump does not resemble a real long jump at all.
Hitting the Take-Off Board
It should be obvious from the animation that the process of trying to hit the take-off board is far from realistic in the original Track & Field. A few meters before the board the athlete stops running. Standing on one leg he starts to glide towards the pit at high speed while the player tries to time the press on the jump button at a position as close to the board as possible. Sliding feet at its finest! It plays just like chicken race and actually works very well. The absurd gliding, however, would look even more ridiculous together with the realistic animation of our game. It is possible to use the chicken race principle without a gliding animation, but this is at the expense of the player’s awareness of the position of the foot.
However, in real life hitting the take-off board has nothing to do with timing. Every athlete uses a run-up of a certain number of strides (typically about 20 strides). The position of the take-off foot after the last stride depends on the starting point of the run-up and on the length of the strides. Besides speed the most important aspect of the run-up is therefore consistency. If the athlete runs in a different way from time to time, he will only hit the board by pure luck. (Although it is possible to make small adjustments of the stride length during the final strides in order to hit the board this is accompanied by a loss of speed and is generally not recommended.)
It is no wonder that this aspect has been excluded from previous track and field game as they do not simulate the stride in a realistic way. Due to sliding feet during running you could actually say that a running stride, properly defined, does not even exist in these games. As we on the contrary have come up with a way to portray a stride that increases in length as the speed increases, we should be able to include it.
From a game play perspective it should also make the striving to hit the board more complex and interesting. For a start you have to consider the starting point of the run-up. A very short run-up may not be long enough to reach top speed but the fewer number of strides will not allow a large accumulation of inconsistencies. A run-up of normal length will allow the athlete to reach top speed, but there is a larger risk of missing the board. A run-up is overlong if the athlete is not able to sustain top speed all the way to the board.
You must also be prepared to adjust the starting point. You need to shorten your run-up when your form is bad or as fatigue begins to accumulate as you will then run slower and consequently your stride length will be shorter. The wind will also influence your speed and stride length and erratic wind conditions may force you to adjust your run-up from one attempt to another during the same competition (yes, we intended to include wind in the game).
And finally, as the consistency factor itself comprises the input of several button presses (typically 50-70), it is a much more versatile factor than a single timing press.
To my surprise everything worked perfectly! The feeling was very similar to a real long jump run-up. This was a testament of the quality of the time-velocity curve and our method of simulating the running stride. Though, not as easy to comprehend as the chicken race, the mechanisms become rather intuitive after only a few tries.
The Take-Off Angle
The animation during the take-off in Konami’s Track & Field also indicates that the mechanism behind it is a far cry from the real thing. When the jump button is pressed, the athlete comes to a sudden halt and remains in what seems like a paralyzed state until the button is released. As the time the button is held down provides the input, the frozen phase of the animation is rather hard to get around. Not only would the interruption of animation clash badly with the otherwise fluent and realistic animation, but the nature of the input also feels very ill suited to the extremely fast and highly powerful action of the take-off.
The notion that there is an optimum angle of the take-off is absolutely false. The take-off angle is the resultant vector of the horizontal speed and the vertical speed at take-off. That is, if the vertical speed of two jumps is identical, the jump with the higher horizontal speed will have the lower take-off angle (and be the longer jump) while if the horizontal speed of two jumps is identical, the jump with the higher vertical speed will have the higher take-off angle (and be the longer jump). The best long jumper of all time, Carl Lewis, had a lower take-off angle than most of his competitors as he was the faster runner.
The horizontal speed equals the running speed that has been developed in the run-up while the vertical speed is the result of the vertical force that is created during the brief ground support phase before take-off. The time the athlete has at his disposal to achieve the vertical speed is only about 0.15s so naturally the vertical speed can never be of the same magnitude as the horizontal speed. The take-off angle of Carl Lewis was typically below 20° which tells us that the vertical speed was about 1/3 of the horizontal speed. While the potential to obtain vertical speed is not wholly independent of the running speed, the basic principle should still be to maximize both.
Considering this information about the take-off, it would make more sense to exchange the angle factor with a vertical speed factor. After the jump button has been held down for an optimum amount of time, the increase of vertical speed is reversed and the speed thus begins to decrease. However, the problems of the input method have already been mentioned.
A more appropriate control method should involve a test of speed as the take-off is an action of speed and power just like running is. Running is executed by alternate presses on the ‘A’ and ‘B’ buttons on the right side of the screen of the GBA. The direction pad on the left side of the screen could then be used for the take-off. A combination of the ‘Down’ button and the ‘Up’ button seems right. The shorter the time between ‘Down’ and ‘Up’, the higher the vertical speed. Due to the construction of the direction pad, ‘Down’ must be released before ‘Up’ is pressed. This is beneficial as it increases the time span and as a consequence the differentiation of the outcomes. The down & up aspects of the button combination is also very fitting as the athlete actually lower his center of gravity in the stride preceding the take-off to allow for a larger vertical lift. We decided to allow a single ‘Up’ press to result in a jump with a minimal vertical component. Though, this rudimentary way of jumping will not result in world class performance, it is still possible to achieve rather well. It will allow beginners to make successful jumps from the start and may actually also be used by more experienced player as a ‘safety jump’.
Without an adherent timing requirement, the ‘Up’ and ‘Down’ combo may be more or less out of phase with the animation. Additionally, as hitting the board no longer depends on timing skill, the inclusion of a new timing factor is certainly welcome. Preferably, the ‘Up’ press should be timed with the moment the take-off foot reaches the ground just before take-off. As there is always a certain loss of running speed during take-off, I decided to let the quality of the timing determine the magnitude of the horizontal speed loss. This would correspond rather well with the situation of a real long jump. If the foot reaches the ground prematurely, its position will be too far forward of the center of gravity of the body and consequently a breaking action and loss of horizontal speed will be the result.
Remember that both stride frequency and stride length increase with speed. High stride frequency obviously makes it more difficult to time a certain moment of the stride and in the long jump it consequently becomes notably harder to time the ‘Up’ press at higher running speed. This kind of negative feedback also exists in real jumping. High running speed naturally makes the technical execution more difficult and also puts more demands on the muscle at the take-off.
In my opinion it is very positive that the timing skill will involve the player to concentrate on the animation of the athlete. There is also a rhythmical aspect as a sound is heard every time the foot touches the ground. In other track & field games meters of different kinds are introduced to provide for timed presses and this undoubtedly weakens the link between the animated athlete and the player.
We made use of yet another factor of the long jump that I have not seen in any other track & field game. The wind can have a major effect on the results in several track & field events. An average tailwind of 2.0m/s will reduce the time in 100m by about 0.15s. In the case of a race, the athletes have no control whatsoever on the wind. In the long jump, however, an athlete has one minute at his disposal to begin the run-up. The athlete can take advantage of this to wait for an advantageous wind. There is of course no guarantee that the wind will get better. It may very well become worse. So this introduces a bit of hazard to long jumping and considering the other strictly skill-based factors, a gamble factor should only spice up the game. It was not hard to implement into the game either. At the start of the run-up, a clock that displays the remaining time and an animated streamer that indicates the wind were placed and then we also had to simulate several different wind conditions. However, we made the wind factor optional. As I have already noted the wind will affect the stride length and there is a risk that the complexity of the run-up will overwhelm the first time player, especially as the stride length dependant run-up has not been used in games before.
A New Model
We can now make a model of the factors involved in the long jump of Fila Decathlon.
Compared to the previous model, the added factors make the event much harder to master. However, considering the nature of the factors and the fact that two factors are optional (adjustment to wind conditions and vertical speed), it is not difficult to make a successful jump (although at a low level). This is basically what we wanted to achieve. A novice player will make rapid and steady progress at first but after a while progressions come at a much slower rate. And importantly, even at a very advanced level the player will feel that it is possible to jump justa little bit longer.
When I designed the remaining eight events I reasoned in a very similar way. A real life model of the factors that influence the result constitutes the basis. The factors are then considered from a game play perspective and different input methods are tried and tested. It is noteworthy that the real model almost certainly can provide interesting game mechanics. The problem is often to be able to simulate underlying aspects of the models or its factors in such a way that those game mechanics transfer well into the simplified reality of the game. Our simulation of the running stride illustrates this. Without it, several factor of the real life model would not have been possible to use. Not surprisingly, the simulation of the running stride also came to good use in other events like high jump, pole vault and javelin throw.
How then, did the public and the press receive Fila Decathlon? We were aware of the risk of trying to change the cut-in-stone formula of previous track and field games. Certainly it will frustrate players who expect world class performance from the beginning. We expected that those who took the time to play it thoroughly would appreciate its depth and attention to details.
We were more or less right though we didn’t expect so many professional reviewers to be downright bad at the game. We got a lot of good reviews and it was obvious that the reviewers in these cases were rather good at the game as they did not complain about the difficulty level. We also got a lot of so-so reviews. In all those reviews the reviewer listed the difficulty level as the main problem. Very few reviewers actually slated the game and those who did apparently never managed to place anywhere else than last irrespective of the event. There seemed to be a very strong correlation between the skill of the reviewer and his/her appreciation of the game
Generally, the readers/buyers opinions in connection to reviews/sales were very positive. I would like to quote one buyers opinion at Amazon.co.uk: “It's not just like your bog-standard International Track & Field, button-bashers; It takes more skill and good timing. Trying to beat all world records and your personal bests will occupy you forever. Unlike other games you can get for the GBA, this game does not get boring when you've played it for 24 hours non-stop”. This reaction is exactly what we worked so hard for to achieve. But again, there were those who thought that the game was too difficult and failed to appreciate it.
The difficulty of the game was often related to complicated button controls. While we expected such criticism, we still think it is a bit unfair. Even if more complicated than other track and field games, they are always logical and intuitive. A jump is for example performed by a ‘Down’ and ‘Up’ press in both long jump and high jump. There are numerous games that have a lot more complicated button controls. Compared to many fighting games, the button combos of Fila Decathlon are ridiculously simple. We suspect that tradition has a big part in this – track & field games are supposed to be pick-up-and-play games (and Fila Decathlon could still be described as one, but only after the initial threshold has been passed).
Interestingly, a friend of mine with very little videogame experience has achieved the best (to my knowledge) results in Fila Decathlon. He was one of the athletes that I filmed for the animation of the game and had the opportunity to test the game during development. It is fair to say that he was absolutely awful at it. He was later given the finished game and begun to play it obsessively at home and started to experiment with different innovative techniques and external tools of assistance (this alone and the problems associated with it would make up an article of its own). He has been able to totally destroy the best computer controlled competitors and has to this day run 100m at 9.55 and scored above 10.000 points in the decathlon. This also illustrates the impossibility of keeping the result level at a realistic level. If the parameters were to be adjusted to his performance, the less skillful players would have been left even more behind.