In this paper, "Spiral Steps", whole 3D information visualization, is proposed for data having some attributes. In the method each element is represented by an icon, which represents its attribute as color and shape. Grouping and sorting one of the attributes, we allocate all group of the elements spirally and hierarchically. The Spiral Steps can be regarded as a revised version of "Circular City", proposed by the authors previously. By the Spiral Steps we can obtain better visibility of elements and apply the visualization to non-structured data as well as structured data.

View full abstract

The hands are the most characteristic part of the human body from the standpoint that they are related to other elements through actions like grasping and touching. Several elaborate animation techniques for the hand with 3DCG have recently been proposed, but they are not good enough for simulations of hand touching, which requires simulation of complicated actions and position-dependent elastic forces in the hand. In this report, an algorithm is proposed which the user can operate intuitively for the deformation and the color change of the hand induced by touching. An outline of the algorithm is as follows; 1) elastic deformation, depending on the position of the hands, is controlled by channel images, and 2) color change is expressed by synthesizing three kinds of texture images.

View full abstract

"theStrings the Collection" is an interactive media-art work, which consists of three versions from each different viewpoint. The theme is "figure of sound" and "theStrings the Collection" expresses "figure of sound" based on each motif as an attractive real-time CG by inputted sound into a computer. One motif is synchronized swimming and Strings dance synchronized with sound. Two is echo and Strings create a sound tree echoing sound. Three is jewel and Strings train a sound jewel by sound. This paper is written about relations of sound and shape and motion in theStrings.

View full abstract

People usually try to show themselves attractively, since attractive looks are generally preferred. We propose a new method for modifying facial data in order to improve facial beauty while retaining the original features as much as possible. Based on investigations into related work of facial attractiveness,the targets for modification in our method are small and⁄or unbalanced eyes, and a projecting mouth. We apply these modifications to 3D facial data sets. The data sets have the characteristic points that are classified into facial parts. The coordinates of the characteristic points are moved, the modification affects the raw data points surrounding the characteristic points, and a modified face is represented. We evaluated the modified faces of eight people with a questionnaire survey. As a result, we confirmed our method was useful for increasing the beauty of facial data while retaining the original features as much as possible.

View full abstract

To date, more high resolution 3D animated models are required to present important details and fine structures, however, sometimes such high resolution models are un-necessary and undesired,especially on web and mobile phone environments. Though there are many well-known algorithms dealing well on simplifying 3D models, most of them are limited to static ones. Applying these mesh simplification methods to 3D animated models, a good simplified model in a specified pose can be obtained. However, some features of the original animated model, which can be shown in other poses, may be destroyed. In this paper, we propose an automatic method to simplify a 3D animated model which takes the features shown in every poses into account and preserves the geometry details of them. Therefore, a progressive 3D animated model can be generated for mobile or web uses.

View full abstract

To produce four-dimensional (4-D) origami, we fold a solid material along flat planes in a 4-D space. A 4-D space has a fourth axis that is perpendicular to a three-dimensional space. Because a computer graphics (CG) image of a 4-D origami must be drawn on a three-dimensional screen to visualize it, we will produce its CG image using a 4-D painter′s algorithm with a stereogram. First, we will show how to fold a solid material in a 4-D space. After defining front and a back sides of this solid in a 4-D space, we will make mountain and valley folds and thereby produce a "4-D Noshi" CG image. Secondly, we will show how to fold a regular tetrahedron flat by bisectors of its dihedral angles and make a 4-D bird base from a double tetrahedron. Finally, we will produce CG images of this 4-D bird with opened wings using a stereogram.

View full abstract

We have developed a simulation program that presents the process of fastening band-shaped cloth objects. These objects are twisted in 3D space to mimic the movements of tying garments such as Kimono sash. Even using extensive collision handling, this simulation is tough. The difficulty lies in the complex object contacts: The movements of different parts of a long contiguous deformable object conflict with each other causing numerous collisions during fastening process; Such collisions cause implausible deformations; Moreover, the collisions in different angles are prone to fatal penetrations that are not simple to remove. We propose three schemes to solve these problems. One is a pre-step mechanical scheme for calculating fastening force to move the object smoothly. Other two are a post-step geometrical correction scheme to smooth deformations and a set of schemes to remove penetrations. These schemes proved to enable robust and visually realistic simulations with low cost.

View full abstract

In order to design industrial products fitting well to the human body, it is necessary to make full use of information of human anatomical landmarks. These landmarks however cannot be extracted automatically from point cloud data of human body surface obtained by a 3D scanner. In our previous report, we proposed the method of producing foot model automatically by deforming the template of known landmark positions using Free Form Deformation (FFD) method. In this method, it is possible to estimate positions of the landmark by deforming the template model with the anatomical feature retained. However, estimation error in the method was more than permissible amount. This paper discusses an analysis of the factors of estimation error, and proposes revised method of reduced estimated errors for landmarks around the ankle.

View full abstract