Motion Laboratory

Motion data have been widely used for recent digital content, but such data are generally intended only for human body motions. In this study, I attempted to reproduce the folding of a fan with human body motion as CG animation by using a motion-capture system. The motions of a fan, with markers attached at two points on the left and right outer ribs, were simultaneously captured with human body motions by an optical motion- capture system. First, the pivot position of the folding fan and the angle between the outer ribs are calculated from these marker positions. Next, the angle between the outer ribs is divided into 16 parts to reproduce the opening and closing motions of the folding fan, and then 15 tip positions of the folding fan are calculated as mountain or valley points. Sixteen polygonal surfaces are each created using the adjoining positions of one mountain point, one valley point, and the pivot. The reproduced motions of a folding fan are simultaneously displayed with the human body motion acquired by motion capture. To confirm the results of the CG reproduction of a folding fan, an evaluation experiment was conducted with ten students. The results suggest that the motion was well reproduced and that it would be interesting to use this motion for various content.

The purpose of this study is to raise interest in Sugoroku, a kind of Japanese board game of the Edo period, and to support exhibits of it at museums. I developed a system for playing Edo-era Sugoroku using modern interactive techniques. In this system, the user rolls a dice-type device equipped with a microcomputer. By acquiring the value from the triaxial acceleration sensor on the dice-type device, the system detects when the dice starts rolling and judges which face (1 to 6) is facing up when the dice stops. Furthermore, by projecting the game's progress on the board with a projector, the system shows the current position of the players and the next candidate destinations. In addition, LEDs are attached to the dice to light it up in an arbitrary color and to indicate the next player's turn. By adding these elements, it is possible to provide a whole new way of enjoying Sugoroku as it was played in the Edo era. An experiment was conducted to evaluate the operability of this system and to measure users' interest in Sugoroku of the Edo period. As a result, 80% answered it was easy to understand Sugoroku using the system, and the added elements were also interesting. In addition, it was confirmed that the dice-type device is easy to use and thus has high operability. Therefore, the proposed system is considered a useful tool for displaying a traditional game.

With the development of 3DCG, it has become possible to accurately record and reproduce traditional performing arts and outdoor events, which have been difficult to digitize so far. In this research, I developed a system that edits walking paths and simulates traditional processions by 3DCG for events called Nerikuyo, which are difficult to record and reproduce by video. In this system, it is possible to set a route by clicking the position on a map where the procession passes and to simulate walking through the created route using 3DCG. The landscape was reproduced in 3D by importing an omnidirectional image and 3DCG models. Since Nerikuyo is sometimes performed even in places with elevation differences, such as stairs and slopes, whether it is possible to move due to the difference in elevation is determined, and if a route where movement is impossible is inputted, the system selects the bypass route with the shortest distance. The system was experienced by 10 students, and they evaluated it using a questionnaire. As a result, 70% of the students answered that this system was effective for learning about the composition of Nerikuyo.

Maki-e fountain pens become degraded by light and humidity, so they are not suitable for permanent exhibitions. There is also the problem of some fine patterns being difficult to see with the naked eye. In this research, I developed an exhibition-support system for fountain pens using a head-mounted display(HMD) and a pen-type device. The purpose of this system is to allow museum visitors to view Maki-e fountain pens more freely. This system allows users to preview 3DCG of fountain pens displayed in VR space using the HMD while operating the pen-type device. In order to intuitively operate the 3DCG of the fountain pen, the 3DCG is rotated based on the angular velocity detected with a gyroscope sensor mounted on the pen-type device. In addition, in order to select one from multiple fountain pens with the HMD and to switch among them easily, the system allows users to select a fountain pen to view using only head motion by displaying the cursor at the center of the HMD screen. I asked visitors to actually experience this system at a school festival and conducted an evaluation experiment. From the results of a questionnaire, approximately 90% of visitors said they were satisfied with their experience of appreciating the fountain pens using the pen-type device and HMD. Moreover, a high evaluation was obtained of the system's operability. This confirmed that the system is useful as an exhibition-support system for Maki-e fountain pens.

The purpose of this study is to verify the effect of raising the opposite side of the dominant arm of windmill pitching in softball. I measured the windmill pitching of five players by using an optical motion-capture system. The speed of the dominant hand, twisting of the wrist, speed of leg drawing on the dominant-hand side, and the rotation angle and speed of the waist were calculated as physical feature values. The feature values of normal pitching are compared with those of pitching without raising the opposite side of the dominant arm. As a result of analyzing the calculated feature values, the speed of the dominant arm decreased as expected for a pitching form made without raising the opposite side of the dominant arm. Furthermore, also as expected under this condition, the rotation angle of the waist became narrower and the waist's rotation speed became slower. The speed of leg drawing was expected to become faster by lowering the opposite side of the dominant arm but actually became faster even in the case of a pitching form without raising the opposite side of the dominant arm, which, in addition, increased the speed of the wrist's twisting. From these results, the waist can rotate more openly by raising the opposite side of the dominant arm. Moreover, acceleration of the dominant arm and acceleration of the waist both became stronger by dropping the opposite side of the dominant arm after raising it.

In optical motion captures, markers are attached to the surface of the human body, and thus the positions of the markers are different from the actual joint positions. Therefore, this paper investigates whether accurate feature values can be calculated using only the marker points. Targeting a tiptoe movement supported by one leg, the so-called arabesque, as one of the classic ballet motions, the joint positions were calculated from the acquired motion data. Then, the difference between the feature values using these calculated positions and the feature values obtained by the marker points were compared. In both cases, the angle of the right knee was small when the angle of the crotch was large, and vice-versa. Therefore, it was found that the angle of the right knee and the angle of the crotch could be calculated using only marker points with almost the same feature values as those far calculation using the joint positions. However, the horizontal distance by the joint positions was shorter than that by the marker points. That is, the horizontal distance between the toe and the center of gravity was found to be highly accurate when using the joint positions.

In recent years, it has become possible to view valuable materials on a screen using 3DCG and VR technologies, but it takes time and effort to make many materials 3D. In this research, I aimed to create 3D contents efficiently using existing archives owned by museums and proposed a method to generate 3DCG images of fountain pens and I'm assuming this is your meaning here to reduce their image data for use in VR. First, the curve of the main body shape is created using existing radius data. The data size is reduced by outputting the coordinates only every 10 pixels, or at the point where the sum of the differences of four points among the radius data with continuous rows is from 2 pixels to 10 pixels or less. After that, the curve is rotated 360 degrees around the vertical axis to create the main body shape. After creating the 3DCG shape of the fountain pen, the developed image of the fountain pen is attached to the body shape as a texture. For the shape of the pocket clip, a curve is created corresponding to the main body and rotated by the angle of the clip. The upper and lower parts of the main body without existing data were interpolated by three kinds of shapes, and the thickness of the clip and the protrusion of the lower part were manually created. In this research, 20 of the 44 Maki-e fountain pens in the National Museum of Japanese History were converted to 3DCG. I compared the number of polygons of 3DCG on the fountain pen to check whether this method was able to efficiently reproduce the shape and reduce the image's data size. The proposed method produced only 15% of the number of 3DCG polygons compared with creating 3DCG using all pixel data, and this substantial reduction did not incur any degradation of the object's original shape.

The installation of crime prevention products is very effective in preventing burglary, which is the main method of intrusion and theft. Therefore, I developed a simulation system of crime prevention measures using 3DCG to deeply understand the capabilities of security products and gain an awareness of crime prevention measures in daily life. In this system, a player simulates an experience under the assumption that a player as a burglar enters a house where security products are installed. The user can confirm the need for security products by looking at the position of these products and their performance from the viewpoint of the burglar. Players can select various security products to activate their function and confirm their effect. A time limit of 10 minutes and a risk value indicating the danger level of the criminal were incorporated in the simulation. From these two values, this system shows how much the user was able to suppress the damage using the security products. To evaluate this system, ten people who did not use security products completed a questionnaire and participated in a confirmation test of crime security products. As a result, the scores of both the questionnaire and the confirmation test were high, demonstrating that this system is useful for understanding security products.

The purpose of this research is to support the learning of the Japanese martial arts form called Shorinji Kempo as well as the virtual reproduction of its solo performance in tournaments. Therefore, the author obtained Shorinji Kempo motions by motion capture and then developed a system that could simulate a performance with 3DCG. First, I captured eight basic actions of Shorinji Kempo and two actions of turning using an optical motion-capture system. Users can freely create a performance by combining six actions from the eight basic actions, and then play it as 3DCG animation. In order to apply conventional rules to the created performance, the system restricts the choice of basic actions and the body's direction at the end of the performance. In order to prevent users from repeatedly implementing the same action, the created performance consists of six distinct basic operations, where the number of composed motions and the frequency of each motion are counted when users add a basic action. In order to apply rules requiring that the body both starts and ends in the front direction, a direction-change motion is automatically inserted when these rules are violated. The utility of the system was evaluated by six members of a university's Shorinji Kempo club. As a result, they confirmed that this system could be effectively used to support practice of the sport.

The purpose of this study is to support the creation of contemporary dance choreography using 3D motion data acquired by motion capture. In this study, I have developed "Body-part Motion Synthesis System" (BMSS) to automatically synthesize short choreography motions by using 3D motion data and to create phrases or sequences by combining them. Furthermore, the system simulates these motions in 3DCG. This system runs on a tablet computer, so it is possible to intuitively control it by touch operation. In automatic generation of short choreography motions, this system synthesizes such motions by replacing body-part motions or by mixing whole-body motion. The generation of unnatural motions is prevented by adjustment of synthetic timing and transition time based on the center of gravity position, the ground state of the feet, arms, or torso, and the difference of pose between before and after synthesizing. In its automatic generation of phrases, this system evaluates connectivity between created or saved short choreography motions using the similarity of pose included in each choreography motion, which is determined by the distance between poses. Based on this, a few choreography motions that contain a pose similar to the pose of the choreographic motion entered by user are connected. In addition, this system randomly selects a connecting motion from within the range of a threshold in order to make variations from the results.To evaluate the automatic synthesizing system of choreography motion, an experiment was conducted with 25 students who study dance in Japan, USA, and UK. From the results of the experiment, the effectiveness of this system for supporting dance creation and understanding of motions was confirmed regardless of country or dance experience. A choreography-creation experiment was also conducted with the participation of four professional choreographers. The results clarify the situations and problems for which the system could work effectively. Next, an experiment was conducted to evaluate the effectiveness of the method of handling phrases. In this experiment, subjects were eight students who study dance choreography at university. The results show that automatically synthesized phrases were effective for creating a new idea of motions and supporting dance training. Consequently, this system's method of automatically synthesizing phrases was confirmed to be effective for supporting dance creation.

• Yuho Yazaki, Asako Soga, Bin Umino, Motoko Hirayama, Automatic Composition by Body-part Motion Synthesis for Supporting Dance Creation, Proc. of International Conference on Cyberworlds 2015, pp.200-203 (Gotland, Sweden), Oct. 2015