Non-Conventional Interaction Study on Rythm Games

Transcription

1 Non-Conventional Interaction Study on Rythm Games Márcio Zacarias and Luciana Nedel Instituto de Informtica Universidade Federal do Rio Grande do Sul Porto Alegre, Brasil Abstract The video game industry is experiencing a very positive moment. High quality and low cost hardware allows the fast popularization of game consoles and the games culture dissemination for a diverse group of gamers, including from addicted to casual players. However, even if we have new games with high quality rendering and involving stories, comfortable and innovative interfaces are not yet a reality everywhere. In this paper we explore the use of nonconventional interfaces for casual games in portable computers, a platform that has being neglected in terms of user interfaces for games. We describe a study of unusual interaction techniques and devices with rhythm games. Several interaction possibilities were implemented and tested with users, most of them involving the Nintendo Wii devices, a low cost solution. Our experience on developing and testing it with users is presented and discussed in this paper. Finally, we demonstrate that the use of creative interactive methods is essential to achieve a complete user experience for casual games in terms of efficiency, fun, and comfort. 1. Introduction The video game industry is experiencing a fantastic growing in the last few years. It was considered the more profitable entertainment activity in England overcoming cinema and music. Additionally, games industry is also growing in diversity, presenting each year a larger list of types and titles. With new types of games, new needs as the search for innovative interaction techniques should be addressed. In this context, personal computers are yet an important platform, comparable to last generation game consoles regarding the graphics power. However, unfortunately we cannot say the same regarding interaction. While the industry of game consoles is becoming to explore and produce alternative interaction possibilities [4] [8], video games for personal computers are yet based on mouse and keyboard interaction, that are very efficient for conventional 2D applications, as well as, for some kinds of games, like the First Person Shooting games (FPS), for example. Gamepads, as the ones used on game consoles, are alternatives to keyboard and mouse and provide a good usability for some types of games but are far from an universal solution that gives an enjoyable experience for many groups of gamers. Examples of games that require more creative interactions include the casual and party games, usually in the preference of users that do not have much time to invest playing games, or that use games as a socialization tool. Casual gamers were the main target public of Nintendo Wii and Nintendo DS consoles. Both have introduced new interaction metaphors, achieving an enormous success in the video games market since A sub-set of casual games is the rhythm games, where music plays a fundamental role on the gameplay. In this kind of games, the game history is normally very simple, focusing mainly on the interaction experience. Probably the best example of a rhythm game is the Guitar Hero, distributed by Activision Inc. In this game, the user is challenged to hit the musical notes that appear in the display, having the experience of a guitar player. The game is simple and can be played by users with different levels of expertise but, undoubtedly, its success is due to the new interaction devices that were proposed, simulating simplified guitars and, more recently, other music instruments. Due to the success of Guitar Hero, other games in the same genre were released and explore similar interaction metaphors. Examples include Dance Dance Revolution [1], a pioneer on rhythm and dance video game genre, where players stand on a dance platform and hit colored arrows with their feet to musical cues. Osu! Takakae! Ouendan! [6], for Nintendo DS, is a rhythm game where the player uses the stylus to hit circles that appear in the screen also using musical cues. The last Nintendo consoles (Wii and DS) introduced innovative interaction devices, increasing the user experience 163

2 and extending the playability of the new games. Unfortunately, these technological advances are not yet commercially available for personal computer, motivating this work. In this article we describe a study on using nonconventional interaction techniques and devices to play a rhythm game on a personal computer platform. We implemented and tested several different interaction possibilities involving Nintendo Wii devices, as well as the Bratrack system a motion capture system based on two infrared cameras, and the conventional mouse, keyboard and gamepad. Our experience on developing and testing it with users is presented and discussed in this paper. The main contributions of this work includes: a set of new interaction methods for rhythm games a low cost version of a motion tracking system using the Wiimote device a comparative user study of the methods proposed, including statistical analysis Additionally, the following assumptions are demonstrated: non-conventional interaction techniques are as efficient as the conventional ones, but more pleasing interactions involving the whole body are as comfortable as the other ones studied here the previous experience of the users with games or new non-conventional devices is relevant but not proportional to the entertainment experienced Finally, we would like to demonstrate that the use of creative interactive methods is essential to achieve a complete user experience for casual games in terms of efficiency, fun, and comfort. Furthermore, it can be achieved with low cost solutions. This paper is organized as follows. Section 2 presents some basic concepts and previous work of the area. Section 3 overviews the rhythm game MoonBunny used to support the interaction tests. In Section 4 we detail the techniques we conceived and implemented to interact with the game. In Section 5 we present the user studies carried out, and in 6 an analysis of the results achieved. In Section 7 we present our conclusions and future works. 2. Related Work Since the introduction of the commercial video game consoles in the 70s, interaction devices and graphics realism has been persecuted issues. The first video game console (the Magnavox Odyssey) had a game controller composed by a button and a dial, for 1D movements. The second one, the Atari 2600, introduced a 2D controller, with a button and a joystick. Then, in 1983, the Nintendo Famicon console introduced the first gamepad, allowing more complex interaction with two hands. This innovative game controller was still used for 2D interaction, but included directional buttons for 2D movements (instead of the joystick) and 4 other buttons for general purposes. Since then, the gamepad became the standard for video games interaction and is still in use by the most part of the consoles commercially available nowadays. Gradually, its complexity was increased, allowing more expressivity for the users. As a consequence, games were each time more difficult to master, restricting the new titles to hard-core players. In 2006, with the released of Nintendo Wii console, this scenario changed. The industry finally introduced a new interaction metaphor where users play games by moving themselves in the space as they are inside the game. The great success of this new console is thanks to this new interface concept, allied to the correct exploration of it for casual games, and a very convenient price. The innovation introduced by the new Nintendo Wii, quickly provoked the interest of programmers and motivated the exploration of these new devices for other purposes and in other platforms. One of the first and most popular initiatives in this sense is the work of Johnny Lee [2], published at first on YouTube and then in his Web Site. Johnny Lee developed several small projects on using the Wiimote control to interact with a PC and put the code available over the internet. Some examples are a finger tracking system, a low cost multi-point interactive whiteboard, and a head tracking for desktop VR displays. Schlömer et al. [9] proposed the implementation of gestures recognition metaphors initially developed to be associated with the mouse and the stylus used to interact with touch screens using the accelerometers and infrared camera of the Wiimote. Their system allows the pattern recognition of arbitrary movements made by the users using Markovian networks for training and recognition. Leder et al. [3] have integrated the Wiimote to an existing gesture therapy system of computer simulated therapy exercises. Results obtained indicate that the Wiimote model is promising for integration into clinical and home-based rehabilitation exercise therapy systems. More specifically in the video games area, Schou and Gardner [10] describe a project that integrate the Wiimote, a game engine and a multi-screen virtual reality cinema. The project idea was to create a very immersive experience for the player, allowing the interaction of the user with multiple screens (like in a CAVE) with a Wiimote. Results shown that the system has a high level of immersion and interactivity. 164

3 3. MoonBunny MoonBunny is a rhythm game developed in our research Lab. using Python script language and the game engine Panda3D. MoonBunny is a one-player game where the avatar player the BunnyBoy is a boy with a bunny fantasy. Color rings are spread on the scenario and moves in the direction of the avatar according to a music rhythm. The objective of the player, that is always flying, is to pass inside these rings exactly in the right moment. Figure 1 shows a snapshot of the game, including the avatar and a set of rings. The other signs on the top of the screen are the representation of the buttons to be pressed (magenta squares) and the target (in orange). Figure 2. Ranking and points given for each ring crossed, obtained measuring how far from the ideal timing the player is. Figure 3. Scores and conditions needed to achieve each one. 4. Interaction Modalities Figure 1. Snapshot of the MoonBunny game including the following game elements: the BunnyBoy (player avatar), the current score (on top), color rings, rhythm marker (squares on the top), ranking for the last ring crossed (upper left). In the original version of the game, it can be controlled only with a keyboard or a gamepad. The moment where the player confirms pressing a key or a button he/she is passing inside the ring is very important since the player performance is judged according this timing. More synchronized with the music rhythm the player is, higher will be his/her score (Figure 2 presents the ranking table used). The avatar position in relation to the center of the ring is not considered. The only constraint is that the avatar should pass inside the ring. In the end of the phase, the player receives a final rank (see Figure 3), calculated according to the percentage of ranks obtained along the game, and that synthesize his/her global performance. In order to achieve the objectives of this work to test and evaluate unusual interaction techniques and devices with rhythm games, we proposed and implemented 10 different modalities to interact with the MoonBunny game. Then, all of them were compared and the results tabulated. The modalities used the following devices: keyboard, gamepad, mouse, Wiimote, Nunchuk, balance board [4], and the Bratrack system [7]. The interaction requirements involved the accomplishment of two main tasks. The first one, the avatar movement, consists on making the avatar fly and pass inside the rings correctly. The second one, the rhythm marking while crossing the rings, occurs in the same way of the game Dance Dance Revolution [1], indicating a direction (left, right, up, down). In MoonBunny, however, each of these fours directions is mapped to a color. The player should indicate the direction that corresponds to the same color of the ring he/she is crossing. Concerning navigation, even if Moon- Bunny is a 3D game, it is important to state that the interaction is, in fact, 2D, since the user moves the avatar to left, right, up, or down. We assume that the in-depth navigation of the BunnyBoy is automatic. Following sub-sections explain how these tasks were im- 165

4 plemented in all the ten modalities proposed Keyboard The avatar motion with the keyboard is done using the arrow keys, while the rings marking is done with keys a, s, w, and d used for left, down, up, and right, respectively (each one mapping a different color). We used these keys because it follows the same standard used on first person shooting games (FPS). In this way, we believe it is intuitive for players with previous experience on this kind of games Mouse and keyboard This modality is, in fact, a variation of the first one. The only difference is that the mouse substitutes the keyboard to move the avatar. The rings marking continues being done in the same way. We used the Balance Board as an interface to move the BunnyBoy and the gamepad for marking the rings. For instance, if the player wants to move the avatar to the left, he/she must place the body weight also to his/her left side. The use of the Balance Board requires a calibration step for each player. As the device operation is based on pressure sensors, the output is dependent of the user weight Wiimote ACC The Wiimote is the device that allows the rich interaction provided by Nintendo Wii console. Besides the set of buttons disposed on this one-hand remote control surface, the Wiimote provides movements with 3 degrees of freedom, as shown in Figure 5. While pitch and roll are detected with the accelerometers, to detect yaw the use of an infrared signal emitter is needed Gamepad In this modality we have used a Dual Shock gamepad, the same used with the Sony PlayStation2 console 4. The BunnyBoy movement is made with the left joystick, while the rings marking with the buttons triangle, circle, square and x to indicate up, right, left, and down. It is also a standard for video games. Figure 5. Wiimote activation schema. Figure 4. Dual Shock gamepad Balance Board and gamepad In this modality we decided to test a new device that few users have had previous contact. The Balance Board was released in July, 2007 and is supposed to be used as an accessory of Nintendo Wii. The Wii Balance Board is shaped like a household body scale and contains multiple pressure sensors that are used to determine the center of mass and the weight of the user. The user should stand up over the Balance Board and move his/her body to interact, what seems very intuitive. We call Wiimote ACC, the modality where only the accelerometers are used. In other words, only pitch and yaw are supported. We used roll to move the BunnyBoy to left and right, and pitch to move it up and down. The directional buttons of the controller were used for marking the rings Wiimote and Nunchuk ACC This modality is an extension of the previous one, only adding a new device. The Nunchuk consists of an extension to be connected to the Wiimote. With an analogical joystick and an accelerometer, improves the creativity power of the Wiimote. In this modality, we use the Nunchuk joystick to mark the rings Wiimote IR This modality makes use of the infrared signal detection capacity of the Wiimote. The avatar motion is done in the 166

5 same way of Nintendo Wii games, pointing with the Wiimote to the position on the screen, where he/she desires to move the avatar. In other words, movements are produced with pitch and yaw rotations. The directional buttons of the controller were used for marking the rings Wiimote and Nunchuk IR This modality is an extension of the previous one, using the Nunchuk joystick to mark the rings Bratrack to support 8 new interaction modalities. As shown in Figure 1, the symbols used to indicate the rhythm marks, correspond to the same symbols on the Dual Shock gamepad (as in Figure 4). When we included the new interaction modes, it did not make sense any more. We then changed it by colored arrows, aggregating direction and color information in the same symbol, as shown in Figure 6. The understanding of the rhythm markers is now much more intuitive. We also included in the game, some screens explaining how to interact with each of the 10 possibilities. The screens are shown to the user to guide his/her choice for a specific modality. One of the proposals of this project was to control the BunnyBoy using body movements, as explained in Subsection 4.4. This modality explores another alternative to capture the body movement, in this case, using the movement of the head, instead of changing the center of mass of the entire body. In order to capture the movement of the user head, we used Bratrack system [7], a very accurate motion capture system based on two infrared cameras. The Bratrack system uses a central server that recognizes marks composed by a set of shiny spheres on a 3D environment and send the 3D position on these marks to other computers, using sockets. In the MoonBunny, one of these marks was attached to a cap used by the players while playing the game. In this way, the user controls the BunnyBoy by moving the head for both sides, up and down. For marking the rings, a gamepad is used as in some of the previous modalities presented here WiiBratrack Due to the high quality of the output data generated by the Bratrack system, the result achieved with the previous modality was very good. However, the Bratrack system is expensive to be used in games. Then, we developed an alternative to Bratrack using the Wiimote, which has an embedded infrared camera. In this modality, we put the Wiimote on a tripod behind the player. The player also uses a cap to play, this time with an infrared led behind in such a way that the Wiimote camera could capture the led light. Regarding interaction, the idea is exactly the same explored in Subsection 4.9, the user moves the head to move the BunnyBoy. For marking the rings, a gamepad is used. 5. The Experiment As mentioned before, the interaction studies carried out in this work were based in controlled experiments with users. In this section we describe the experiments in detail. The original MoonBunny presented in Section 3 was conceived to be played using a keyboard, a mouse, or a gamepad. As discussed in Section 4, we extended the game Figure 6. Snapshot of the MoonBunny game with the new rhythm marker (color arrows on the bottom). The experiment was performed in four steps: explanation of the experiment to the tester, instructions of how to play the game to the tester, tests, and qualitative evaluation. The volunteers have tested all of the interaction modalities of Section 4, presented to them in a random order to avoid any bias. In each of the 10 phases, he/she had 30 seconds for training and 30 seconds to play. The training phase could be repeated many times, until the tester feel comfortable to play. Figure 7 shows some snapshots of volunteers testing the MoonBunny. During the playing phase, we logged the following data: 3D position (screen coordinates and instant of the music) of the BunnyBoy and distance to the center of the next ring to cross along the 30 seconds of the game; ranking of each ring and distance of the avatar to its center while passing; final score; final rank; screen distance followed by the avatar during the test. After the playing game phase, the volunteers were in- 167

6 XII Symposium on Virtual and Augmented Reality Natal, RN, Brazil - May 2010 Figure 7. Snapshots of the user tests. vited to fill a form with their impressions about the experience. They evaluated each modality as very bad, bad, normal, good, and very good, given a rank for each of the four criteria: efficiency, intuitiveness, fun, and comfort. They were also invited to compare WiiBratrack and Bratrack, indicating which one was the best or if both were equivalent. A free space for comments was also offered. Our experiments included 22 volunteers and the total time spent with each one of them was of about 40 minutes. Figure 8. Mean global score, mean distance to the center of each ring crossed, and mean distance followed by the avatar over the screen (indicates the amount of movement made by the user) for each interaction modality. 6. Results 6.1. Quantitative results The data obtained with the logs shown some interesting results. Some of them can be found in Figures 8, 9, and 10. The first clear conclusion that we can derive from the results is that the Balance Board is not a good solution for this kind of game. Analyzing the data on Figures 8, 9, and 10, we can easily verify that it is not efficient at all. The main reason for this bad performance comes from the difficulty in controlling the avatar movement. As mentioned in Subsection 4.4, we move the avatar by changing the center of mass of our own body over the BalanceBoard. Even if it is not so difficult to move the BunnyBoy for the left and right hanging the body for both sides, it is very complicated to hang the body forward and backward while the two legs are aligned. And this was the movement attended to move the BunnyBoy up and down. Another evidence of this observa- tion can be seen in Figure 11. The graph shows the mean distance calculated from the avatar to the center of the ring while crossing it, measured along the entire game phase and with all interaction modalities. While all of them present a very similar result (in blue in the graph), distances measured while using the Balance Board (in red) are completely different. Figure 12 shows the mean path followed by the user with the Balance Board and helps to sustain this statement. We can see the difficulty of the users to pass inside the rings, by observing that many rings were lost. Analyzing the tables on Figures 8, 9, and 10 we verify 168

7 Figure 9. Mean rank (in percentage) obtained with each interactive modality. This rank is given each time the avatar pass inside a ring. Figure 11. Mean distances of the avatar to the center of the ring, during the entire phase (30 sec) and for all interaction modalities. The red line indicate the Balance Board behavior. Figure 10. Mean global rank (in percentage) obtained with each interactive modality. This rank is given when the game phase ends. that, considering efficiency, mouse and Balance Board are respectively the best and worst interaction modalities. However, there are also important aspects to observe on the other options. The similarity on the performance (see Figures 8, 9) achieved with Bratrack and WiiBratrack are impressive. Greater differences appeared on Figure 10 where scores S were assigned 18.18% of the time for Bratrack and only 4.54% for WiiBratrack. For the other hand, the score F where never achieved with WiiBratrack, and achieved 4.54% of the time with Bratrack. Remembering that WiiBratrack was proposed as a less expensive version of Bratrack, we concluded that the objective was achieved. We also observed that there is a correlation between the screen distance followed and the general performance of the users, as shown in Figure 8. The Wiimote IR, and Wiimote and Nunchuk IR are exceptions. Analyzing the graphs with the mean path followed by the users with these modalities shown on Figure 13 we can verify that they are similar, what is logic since the movement was exactly the same. But we can also note a pattern on the movement, where the user, after to pass inside a ring, moves the avatar to the middle of the screen, and then goes quickly to the center of the next ring. This data, allied to our observation of the volunteers playing, lead us to conclude that this strange movement emerges of the fact that the interaction with Wiimote Figure 12. Mean path followed by the 22 volunteers along the game phase, while using the BalanceBoard. Color ellipses indicate the position of each ring, gray lines the optimal path, and the yellow line the user path. IR is very easy and efficient. Then, players have time to accomplish their task (the first objective) and dance (having fun, the second objective) in the music rhythm, what is very important in this kind of game. This dance behavior was also perceived in Bratrack and WiiBratrack modalities for the same reasons. Apparently, more natural is the interface, more relaxed and comfortable will be the user, also having more fun, which is fundamental for games. The performance achieved with the modalities using Wiimote ACC was worst than the one using Wiimote IR. This behavior was, in fact, expected, because the interaction with Wiimote ACC is not intuitive at all. In this modality, the user should turn his/her hand (with the Wiimote) doing a roll rotation, what is much less natural than the yaw used 169

8 Figure 14. Comparative evaluation of Bratrack and WiiBratrack. Columns indicate the percentage of users that found: Bratrack better than WiiBratrack; the opposite; both are equivalent. Numbers in green identify values greater than 50% Figure 13. Mean path followed by users along the game phase, while using the Wiimote IR (up) and WiimoteIR and Nunchuk (down). with Wiimote ACC. A curious result was the one involving the use of Nunchuk. When we use only the Wiimote to interact, the BunnyBoy movement and the rhythm marking should be produced using the same controller, what could be difficult since two different commands are expected at the same time and using the same one-hand. Game players usually interact using a gamepad, holding it with both hands. In order to avoid this problem, we decided to introduce the Nunchuk associated to the Wiimote and the performance was worst than using only the Wiimote. We believe that the reason for this is that people does not have the habitude of playing interacting with two hands. Additionally, the Nunchuk was on the non-dominant hand. However, these are just impressions. More experiments should be done to prove this Qualitative results Qualitative data was obtained from a questionnaire answered by the volunteers where they were invited to evaluate each interaction modality according to its efficiency, intuitiveness, fun and comfort. One of our interests was to compare Bratrack with WiiBratrack to verify how good is our low-priced version of Bratrack. Figure 14 shows a table with a comparative evaluation of both. The higher difference concerns efficiency, what is completely comprehensible, given the high accuracy of Bratrack. Putting this data together with the quantitative evaluation, we can conclude that WiiBratrack is completely viable as a substitute of Bratrack for this kind of use. Another important issue we should take into account in interface evaluations with users is the degree of experience one has with each of the modalities tested. The experience was tabulated in the interval [0.0;2.0]. Values greater than 1.0 indicate a normal or good experience, while the ones near zero indicate a weak experience. The values obtained were: 1.76 for the mouse; 1.71 for the keyboard; 1.19 for the gamepad; 0.28 for the Wiimote; 0.19 for the Nunchuk; and 0.0 for the Balance Board. The analysis of these data shows that many of the volunteers have experience with games (due to the level of experience with the gamepad), only few of them know the Wiimote and Nunchuk, and nobody has experience with the Balance Board. In such a way, this profile induce us to think that part of the good results obtained with the mouse in the quantitative analysis are due to this experience. We can also speculate that some training sessions can significantly change some of the results achieved in future tests. The lack of experience with the Balance Board can also explain part of the bad results. The Balance Board introduced a completely new interaction modality based on a pendulum movement of the whole body. It is obviously non evident for people that tried it for the first time. In order to extract information from the rest of the qualitative data, we used the Dual Scaling technique. The Dual Scaling [5] is based on the analysis of principal components and allows the visual analysis of a huge quantity of data types, what is suitable for qualitative data. Unlike the conventional analysis of principal components, Dual Scaling allows the projection of variables of different domains in a common space, facilitating its visual interpretation. The axes of this space represent the obtained solutions, where each solution explains a percentage δ of the information contained in the input data. In this work, the Dual Scaling was applied on a contingency table where each line represents one of the 10 interaction modalities and each column represents an evaluation criterion. Each cell of this table contains the amount of times that the pair modality/criterion was observed. The application of Dual Scaling results in the projection of the lines and columns of the contingency table on the so- 170

9 lution space. In our case, the solution has 19 dimensions. However, as shown in the graph of Figure 15, only the two first dimensions are relevant and explain 85.34% of the results (δ1 = 63.78% and δ2 = 21.56%). For this, in the remaining of this analysis we will show the orthographic projection of the point on the two first dimensions. Figure 16. Interactive modality vs Criterion + Score. Figure 15. Relevance of the solution vs Dimension of the solution. Figure 16 shows the graph generated with the Dual Scaling. In this figure, blue points represent the interaction methods; red diamonds the combination of a criterion with a score; and gray lines indicate the combination of criteria+scores that better explain an interaction modality, considering the Euclidean distance in the 19- D space. Unfortunately, the understanding of this graph depends of the following legend. Interaction modalities are indicated as: K=keyboard; J=gamepad; M=mouse; WA=Wiimote ACC; WI=Wiimote IR; WAN=Wiimote and Nunchuk ACC; WIN=Wiimote and Nunchuk IR; B=Bratrack; WI=WiiBratrack; and BB=Balance Board. The criteria are labeled as: E=efficiency; I=intuitiveness; F=fun; C=comfort. Scores are: VB=very bad; B=bad; N=neutral; G=good; VG=very good. It is also important to state that criteria labels should be combined with scores (e.g. FVB=fun + very bad). The groups formed in the graph represent the final qualification of an interaction modality for each criterion. For example, the Balance Board (BB) in the right back of Figure 16 is near all of the criteria with a very bad (VB) judgment. Analyzing the graph on Figure 16, we can notice that similar methods are near of each other (WAN and WA, WIN and WI, WB and B) suggesting that, according to the subjects perception, they are not so different. Literally, subjects do not think the Wiimote ACC modality is so different than the Wiimote ACC with Nunchuk. The same is also valid for the Wiimote IR as well as for Bratrack and WiiBratrack. To make the interpretation of the data easier we generated the table shown in Figure 17, where scores are attributed for each modality and each criterion. Figure 17. Summary of the Dual Scan evaluation. From the data tabulated in Figure 17 we generated the graph in Figure 18. For each score on Figure 17 we attributed a value from 0 for very bad till 4 for very good, and the sum of these values for each interaction modality determined the ranking shown in Figure 18. From Figure 18 we concluded that Bratrack was the interaction modality preferred by the users and that the modalities WiiBratrack, Wiimote IR and Nunchuk, and Wiimote IR are the three in second place. From these data we can prove that users prefer non-conventional interaction modalities. 7. Conclusions In this paper we presented a study about the influence of non-conventional interaction techniques in casual games, 171

10 for their kind participation in the experiments. References Figure 18. Global score of each interaction modality according to a generalization of the oppinions of the users. more specifically, in rhythm games. Ten different modalities were proposed to interact with a rhythm game and user tests were done with 22 volunteers. During the development of this project, other informal tests were also made and were not described in this text. However, the feedback obtained with these preliminary tests gave us experience and feeling that were crucial to guide the choice for these ten modalities. From the analysis of the results, it is easy to conclude how important is to provide creative and innovative interaction for rhythm games. Conventional interaction devices and methods as mouse, keyboard and even the gamepad give a good support for players and help them in achieving high scores. However, are far from the rates obtained by non-conventional methods intuitiveness and fun. Casual players looking for some fun frequently play this kind of game. For this public, fun is much more important than performance. As future work, we intend to choose among the interaction modalities the ones that achieved the best rates in this first evaluation and address the collected data in more depth. As mentioned in Section 6, we believe the good rates obtained with the mouse are due to the familiarity of the user with this device. In this sense, we intend to repeat the tests with volunteers that are so familiar to non-conventional interfaces as to the mouse. To achieve our goal, we will probably need to train a group of volunteer, before to carry out the tests. [1] Dance Dance Revolution. dance revolution, [2] Johnny Chung Lee [3] R. Leder, G. Azcarate, R. Savage, S. Savage, L. Sucar, D. Reinkensmeyer, C. Toxtli, E. Roth, and A. Molina. Nintendo wii remote for computer simulated arm and wrist therapy in stroke survivors with upper extremity hemipariesis. In Proceedings of Virtual Rehabilitation 2008, pages IEEE, [4] Nintendo Wii [5] S. Nishisato. Optimal scaling of paired comparison and rank order data: An alternative to guttman s formulation. Psychometrika, 43(2): , June [6] Osu! Tatakae! Ouendan. tatakae! ouendan, [7] F. Pinto, A. Buaes, D. Francio, A. Binotto, and P. Santos. Bratrack: a low-cost marker-based optical stereo tracking system. In SIGGRAPH 08: ACM SIGGRAPH 2008 posters, pages 1 1, New York, NY, USA, ACM. [8] Project Natal [9] T. Schlömer, B. Poppinga, N. Henze, and S. Boll. Gesture recognition with a wii controller. In TEI 08: Proceedings of the 2nd international conference on Tangible and embedded interaction, pages 11 14, New York, NY, USA, ACM. [10] T. Schou and H. J. Gardner. A wii remote, a game engine, five sensor bars and a virtual reality theatre. In OZCHI 07: Proceedings of the 19th Australasian conference on Computer-Human Interaction, pages , New York, NY, USA, ACM. Acknowledgements The authors would like to thank CNPq-Brazil through project number / and Microsoft Brazil for partially supporting this work, as well as all the volunteers 172