This paper presents a new pseudo 3D representation and its rendering algorithm for cartoon-like illustrations, which tolerates viewpoint/view-angle changes. In our representation, target 3D objects are represented as outlines only, not as full 3D geometries such as polygons. These outlines of the illustration are located in 3D space, not on 2D plane. We call this 3D curve as 3D outline. Although this model has restricted 3D information, we can view and rotate the illustration as if there is a full 3D geometry. We have implemented a prototype system, SilF, for viewing and editing 3D outlines with 2D-like sketching interface using common 2D input devices.

View full abstract

We present a system for quickly and easily designing three-dimensional (3D) models of botanical trees using freehand sketches and additional example-based editing operations. The system generates a 3D geometry from a two-dimensional (2D) sketch using the assumption that trees spread their branches so that the distances between the branches are as large as possible. The user can apply additional gesture-based editing operations such as adding, cutting, and erasing branches. Our system also supports example-based editing modes in which many branches and leaves are generated by using a manually designed tree as an example. User experience demonstrates that our interface lets novices design a variety of reasonably natural-looking trees interactively and quickly.

View full abstract

In this report, We evaluated errors of isosurfaces from volume data sampled on face-centered cubic(FCC) lattice. We had previously developed a method for FCC based isosurface generation and modified it. To evaluate the error, we generated isosurfaces using the FCC method and the Modified FCC(MFCC) method and commonly used Marching Cubes(MC) method from the same volume data algebraically defined, then we compared the errors from a true surface among the three methods. We also tried to apply the FCC methods to a medical CT data. As a result, It was shown that the MFCC method made the root-mean-square(RMS) error smaller than that obtained when the MC method was used.

View full abstract

A lot of Non-Photorealistic Rendering methods have been proposed for creating artistic images from a 2D image. In this paper, we propose a method for creating colored-pencil drawings. Although colored-pencil drawings are drawn with limited number of pencils, we can express a lot of colors and gentle textures by changing the strength when drawing strokes and by overlapping strokes of different colors. In order to realize these features, we first determine the colors and the strengths to draw strokes, then draw several types of strokes, such as strokes for outlines, basecoats, and shading, allowing the strokes to overlap each other. When we create strokes for shading, we make their directions to align along the boundaries of regions, resulting in images that are more like the drawings made by artists.

View full abstract

This paper proposes a method for retrieving motion clips using key postures. This method generates a self-organizing map (SOM) from motion files to be retrieved, and then extracts desirable frame segments by specifying postures corresponding each node. Intuitive interfaces are constructed by supplying icons of representative postures that are determined by using a clustering of postures distributed on the map.

View full abstract

Tetrahedral Interpolation meshes which consist of a set of tetrahedrons are required to interpolate 3D meshes with low distortion. However, it is hard to construct tetrahedral interpolation meshes from interpolation meshes, because a number of interpolation meshes is generally huge. This paper presents a method which incorporate approximate rigid interpolation method with multiresolution interpolation meshes. This method constructs tetrahedral interpolation meshes from coarse level multiresolution interpolation meshes and applies approximate rigid interpolation to them. Other vertices are calculated based on refinement of multiresolution interpolation meshes. Our method is absolutely an approximate method. However, our method has an advantage which is easy to construct tetrahedral interpolation meshes, because our method can control precision of interpolation meshes.

View full abstract

We propose an efficient image-based anisotropic reflectance modeling method of textile fabrics based the microfacet surface geometry determined by the cross-sectional shape of fibers, twist of strings, and a type of weave. We first examine the relationship between the reflectance properties and the microfacet surface geometry of a textile fabric using a silk-like synthesized fabric. We then develop an anisotropic model, which provides a bi-directional reflectance distribution function (BRDF) from small numbers of images observing the orthogonal characteristics of textile fabrics, based on KES (Kawabata's Evaluation system for fabrics) method.

View full abstract

In this paper we propose a method to rendering a virtual copperplate print image based on a physical model as a method with non-photorealistic rendering. In particular we pick out the drypoint from many existing copperplate printing techniques. A drypoint print has a remarkable feature that ridged burrs are arising out of the copperplate and the printed lines are slightly ragged and fuzzy. Our method consists of four sub-processes, which simulates the physical actions and phenomena observed in the actual copperplate printing. We indicate effects and availability of proposed method to synthesize a print image with the feature of drypotin by some experiments.

View full abstract

In this paper we describe a technique of camera control (viewdirection and zoom) for high-resolution volume intersection. While there have been studies on camera viewpoints for 3D reconstruction, we in particular focus on human body as the target object. The problem to be solved is rather complex because of the complexity of the shape of human body as well as possible self-occlusion. Thus, it is necessary to take not only the position of cameras but the visibility of the target into account. In order to deal with the problem, we propose to utilize a 3-D articulated model of human body for estimating an error distribution in reconstruction. We then propose an optimization method for determining the camera control on the basis of the error distribution while employing the branch-and-bound algorithm. Experimental results through simulation demonstrate the effectiveness of the proposed method.

View full abstract

In this paper we propose a method to decompose an image overlapped with watercolors to each mono-color paint image. Ukiyo-e: a Japanese traditional multicolor woodblock print, has a mixing color expression which is created by printing with several woodblocks. We have proposed a particle density model for expressing this phenomenon. The approach of this method is to solve an inverse problem, that is, decomposing a mixed color into each basic color element. Our purpose of the method is to estimate woodblocks accurately from Ukiyo-e using the method. As the first stage of this study, we focus on a situation using two kind of watercolors, we study an availability of the method by experiments of actual images.

View full abstract

[in Japanese]

View full abstract

We propose a management technique for a large number of motion data using an object-orientated model and motion frames. In addition, we develop a system for generating animation from an input text that is written by natural language.

View full abstract

We present a user interface that easily models flowers from freehand sketches. The user creates the models by sketching on a transparent canvas displayed in 3D space. To model the stalk of a flower, the user draws two strokes, which indicate a shadow and the appearance of the stalk. To model a leaf or petal, the user first draws the strokes representing the outline of the leaf, and the interface creates a plane-like leaf object. The user then draws modifying strokes representing the 3D form of the leaf, and it transforms the leaf object accordingly.

View full abstract

Keeping the recent trend in mechanization of apparel industry in our mind ,we have started to develop a novel interactive blouse design support system for students of costumery. The salient feature of the system lies in its re-revolutionary Interactive Genetic Algorithms(IGA),which is adapted to select desirable combination of detail design of blouse in a spiral manner with automatic drawing function for efficient pattern making subprocess. In this paper,we present the re-revolutionary IGA interface for design support and automatic drawing function as a part of the initial prototyping of this system.

View full abstract

We propose a general-purposed parallel algorithm for volume segmentation, which does not require any prior knowledge on volume nor region. The algorithm provides binary tree structured split-and-merge mechanism to search and localize boundaries along discontinuities and adapts the partition of volume to those detected discontinuities. This algorithm is independent from order of processing or seed selection and maintains consistencies of adjacent process blocks without their interprocess communication. Our efficient hierarchical step-wised mechanism in merging target evaluation makes merge process so simple and efficient that only two blocks with the same parents are considered at each merge step in binary fashion. Experimental results on an artificial and a CT scan volume data are shown.

View full abstract

Recent advances in volume scanning techniques have made the task of acquiring volume data of 3-D objects easier and more accurate. Since the quantity of such acquired data is generally very large, a volume model capable of compressing data while maintaining a specified accuracy is required. The objective of this work is to construct a multi resolution tetrahedral representation of input volume data. This representation adapts to local field properties while preserving their discontinuities. In this paper, we present an accuracy-based adaptive sampling and reconstruction technique. We have developed a parallel algorithm of adaptive grid generation, according to local field properties and such as orientation and curvature of isosurfaces, until the entire volume has been approximated within a specified level of view-invariant accuracy. Experimental results demonstrate the validity and effectiveness of the proposed approach.

View full abstract