We present a new method to evaluate the impression of paintings, in terms of a mathematical modeling and quantitative assessments of impression in western painting. The present method is based on a detailed modeling of various factors and elements that consist of the composition of the painting. Impressions of 20 subject-people for 50 paintings are measured using the semantic differential method, which are compared to the impressions calculated with the present method. Their correlation is analyzed with Akaike's Information Criterion (AIC) standard. A strong correlation is obtained.

View full abstract

Digital sculpting is a modeling method to make a detailed forming shape like sculpture. Recently, digital sculpting tools are becoming prevailing technology on 3DCG contents production. However, most of the sculpting operations are irreversible, because sculpting tools overwrite the information on shape directly. Therefore, the history operation of the undo redo etc. is often achieved by the snap shot of the shape, and the operation is restricted by the limit of the memory capacity. This paper proposes a new digital sculpting system based on the history of stroke input. This system retains the CSG model with an implicit function set. As a result, each operation is able to be defined as reversible command in our method. Moreover, in our method, every command history item is made re-targetable by maintaining stroke-trails and camera information. It allows users to recover any past state or to cancel a command. Our experimental results indicate the effectiveness of the proposed system.

View full abstract

In this paper, we propose a simple method for simulating the generation, growth and motion of bubbles in carbonated water. Bubbles are known to be generated at small air pockets on the surfaces of objects, such as a glass or a stirrer, and on the microscopic dusts in carbonated water. The number of bubbles and their growth rates differ according to the materials of the objects. In addition, the motion of bubbles also differs according to the materials; bubbles slowly glide on plastic and acrylic surfaces whereas such motions cannot be seen on a stainless-steel surface. Our method can represent such differences by controlling the number of generation points (i.e., air pockets) and applying adhesion forces to make bubbles stick on the surfaces. Because bubbles in carbonated water are sufficiently small, our method regards bubbles as non-deformable spherical particles. We simulate the flow in carbonated water using a grid-based method, together with the two-way coupling technique between bubbles and the convective flow. Additionally, the collisions and fusions among bubbles are also handled to reproduce the complex behavior of bubbles.

View full abstract

We propose a method for modeling nonperiodic aggregates composed of arbitrary elements on a two-dimensional plane. The method effectively generates a dense aggregate from the elements to be arranged. To date, a Poisson disk distribution has been used to generate nonperiodic aggregates; however, the approach creates large gaps among the elements. Dense aggregates of arbitrary elements can be generated with a dart-throwing method; however, this approach is time consuming. To reduce the calculation time, we propose a method that fills in elements in the gaps among the already placed elements. To effectively fill these gaps, our method quantifies the gaps and finds additional positions for the elements. Our experiment showed that our method is more effective than the dart-throwing method. Also, to confirm that our method is generic, examples using two- and three-dimensional elements are shown.

View full abstract

Photo-realistic rendering of participating media, such as steam, water, fire, and smoke, is an important research topic in the computer graphics field. We present a fast and physically based method for rendering participating media based on the progressive photon beams method. For this, we propose a new version using adaptive free path sampling technique and applying a new estimation method for light transmittance functions based on stratified sampling, which overcomes the deficiencies in the progressive photon beams method. Our method is able to take into account not only single but also multiple scattering of light inside the participating medium. Using our method, participating media with complex density distributions can be handled, as well as complex lighting conditions. Additionally, we can obtain images which are guaranteed to converge to exact solutions. Moreover, our method is implemented on a GPU, enabling fast computation of the solutions.

View full abstract

In this paper, we propose "GAYAIT": an interactive video and sound art system that handles a large number of video clips. The system has about 25~80 virtual displays in a 3DCG space, and each of them plays video clips simultaneously. In each video clip, a person is talking a message or singing a song. The user can pick up one or two of video clips with gestures. When the user chooses video clips, the size and the sound of the chosen video clips are magnified. As a result, the selected video clips come into focus, and the message or the song of the video clips can be understood clearly. The user can change video clips selection interactively, and could understand the personality of the people in the video. The user can also feel an instant change between a mass and an individual in our system..

View full abstract

In this paper, we present an interactive approach to visual simulation of whirlwind using grid and particles. Higher-resolution simulation can be carried out efficiently by perturbing air flow computed on a rough grid with small-scale turbulence. The grid-based simulation first provides a primary whirlwind flow field through modeling its underlying heat convection. Then, the particle-based simulation computes the transport energy of turbulence based on a two-equation k — ε model, which allows us to capture the turbulence details. The dynamism of whirlwind is conveyed effectively by the interaction with rigid and deformable bodies, such as flag, branch and paper, placed on interactively-controllable trajectories. In addition, a simplified variant of particle-based volume rendering is utilized to delineate changes in the translucency of whirlwind, which reflects the amount of sand particles stirred up from the ground.

View full abstract

Stereoscopic contents, such as 3D movies, Digital Signage and a glasses-free 3D etc. have recently become more available to the general public. Meanwhile, in terms of labor cost, creating stereoscopic contents still costs more money because it requires several manual labor processes such as drawing a depth map. In the creation of stereoscopic contents, depth information plays an important role for achieving a comfortable product by controlling stereoscopic effect and the depth information can be obtained by stereo matching methods with a set of input images. In this paper, we introduce an innovative approach suitable for the creation of stereoscopic contents, which is combining a fast method and a high-accuracy method. Our approach is constructed from the feature of depth perception and results of our questionnaire survey for verifying cognition of binocular disparity on background domain in stereoscopic contents. We also introduce our application for obtaining depth information by using our approach. Our application is interactive and intuitive, thus users can easily adjust parameters and obtain the results visually. Finally, the processing speed is evaluated by speed measurement and comparison with existing stereo matching methods.

View full abstract

A household accounting system should be a "tool for thinking" to create an economical plan for the future. We propose a new household accounting system that visualizes expenditures and incomes through the placement of rectangles whose areas are proportional to the money amount: Thus, the user can arrange, move, cut off, and connect each item and grasp the state of a family budget. The purpose of this system is making a user look back to the balance of payments naturally by acquiring some awareness through interaction with each item. In this paper, we tested the proposed system by the input and arrange experiment and we describe about the result of long term use.

View full abstract

This paper introduces a multi-media performance system called "MoPH" that allows the user to enjoy playing graphics and sounds using an iPad. The computer graphics and sounds are generated based on the movement of "objects" that are subject to physical laws. Users of the MoPH define and interact with the object using the multi-touch functions of an iPad. We describe design concept and system of the MoPH. Through demonstrations and experiments of users, we consider effectiveness of the MoPH and its challenges of the future.

View full abstract

Hairstyle is one of the most important features people use to characterize one's appearance. Since whether a hairstyle is suitable or not is relied on the aesthetic sense of human, mathematical model could not represent the suitability. We focused on the fact that the suitable hairstyle is related to one's shape. This paper proposes a new technique for automatically retrieving a suitable hairstyle from a collection of hairstyle examples through learning the relationship between facial shapes and suitable hairstyles. A method of hair-face image composition utilizing matting technique was also developed to synthesize realistic hairstyle images. A database depending on users' preference was constructed to retrieve more suitable hairstyles. The effectiveness of the proposed technique was validated through evaluation experiments by themselves and the others.

View full abstract