Abstract from Some of My Papers (4/6)
Abstract from Some of My Papers (4/6)
|
Keywords: real-time rendering, GPU, image-based lighting, dynamic scene, spherical Gaussian
|
Keywords: Image Composition, Interactive, Segmentation
|
Keywords : glare, diffraction, spectral power distribution, high dynamic range imaging
|
|
Key words; animation, skinning, deformation, dual quaternions
->
.
|
Key words; animation, skinning, deformation, dual quaternions
|
|
|
|
|
|
Keywords: chaotic neural networks, lighting design, computer graphics, radiosity method,
combinatorial optimization problem, simulated annealing, genetic algorithm.
Keywords :
Image synthesis; hair rendering; scattering
"An Efcient Method for Rendering Underwater
Optical Efects Using Graphics Hardware"
by Kei Iwasaki, Yoshinori Dobashi, Tomoyuki Nishita
Abstract
The display of realistic natural scenes is one of the most important
research areas in computer graphics.
The rendering of water is one of the essential components.
This paper proposes an efficient method for rendering images of scenes
within water.
For underwater scenery, the shafts of light and caustics are attractive and
important elements.
However, computing these effects is difficult and time-consuming since
light refracts when passing through waves.
To address the problem, our method makes use of graphics hardware to
accelerate the computation.
Our method displays the shafts of light by accumulating the intensities of
streaks of light by using
hardware color blending functions. Making use of a Z-buffer and a stencil
buffer accelerates the rendering of caustics.
Moreover, by using a shadow mapping technique, our method can display
shafts of light
and caustics taking account of shadows due to objects.
Additional information
"Volcanic Smoke Animation using CML"
by Ryoichi Mizuno, Yoshinori Dobashi, Tomoyuki Nishita
Abstract
The animation of volcanic smoke is useful for natural disaster simulations,
entertainments, etc. In this paper, we propose a model to generate realistic animations of
the volcanic smoke. The model is designed by taking the eruption magnitude
decided by the eruption velocity and the initial volcanic smoke density, the buoyancy
generated by the difference between the volcanic smoke density and the atmospheric density,
and the decreasing of the volcanic smoke density due to the loss of the
pyroclasts (fragments of magma); i.e., this model is based on the physical dynamics of the
volcanic smoke. In this model, the Navier-Stokes equations are used, and we solve the
equations by using the method of the Coupled Map Lattice (CML) that is an
efficient solver. Hence, in our system, the behavior of the volcanic smoke can be calculated
in practical calculation time, and various shapes of the volcanic smoke can be generated by
only changing some parameters. Therefore, realistic volcanic smoke
animations can be created by our approach efficiently.
Additional information
IIEEJ 2002; Journal of IIEEJ, Vol.31, No.4, pp.468-476, 2002-7 (in Japanese)
"Network-based Walk-through System using Relief Textures as Distant View Billboards"
by Yuichi Koiso, Ken-ichi Amoh, Yoshinori Mochizuki, Tomoyuki Nishita
Abstract
In three-dimensional computer graphics, a set of polygons are generally used to represent a virtual space. In this
situation, if the finer representation for the virtual space is required, the number of polygons constituting the virtual
space is increasing, which gets rendering speed de-creasing. In addition, to download through the Internet such a large
number of polygons cost much communication time. To solve these problems, we use relief textures, textures with depth
information, as distant view billboards in the walk-through system on a network.
Key Words:
Relief Texture, Walk-through System, Billboard, Virtual Space
Additional information
"Optimization Using Chaotic Neural Network and Its Application to Lighting Design"
by Ryuichi Nanba, Mikio Hasegawa, Tomoyuki Nishita, Kazuyuki Aihara
Abstract
We have developed a chaotic neurodynamical searching method for solving
lighting design problems. The goal of this method is to design interior lighting
that satisfies required illuminance distribution. We can obtain accurate illuminance distribution
by using the radiosity method to calculate interreflection of lights.
We formulate the lighting design problem that considers the interreflection of lights
as a combinatorial optimization problem, and construct a chaotic neural network
which searches the optimum solution of the lighting design problem.
The calculated illuminance distribution is visualized using computer graphics.
We compare this optimization method with the conventional neural network with gradient dynamics,
simulated annealing, and the genetic algorithm, and clarify the effectiveness of the proposed method
based on the chaotic neural network.
Keywords and Phrases:
chaotic neural networks, lighting design, computer graphics, radiosity method,
combinatorial optimization problem, simulated annealing, genetic algorithm
Additional information
IIEEJ 1998; Journal of IIEEJ, Vol.27, No.4, pp.287-297
"Free Form Deformation by Detecting the Closest Point on a Curve"
by Tomoyuki Nishita, Ryouji Matsuda, and Kazuhisa Takae
Abstract
In geometric designs (or CAD systems), the interactive process of curves is important. This paper proposes a new calculation algorithm of the closest point on a curve to a point on a screen and its application method to deformation/morphing of images. Even though it is easy to evaluate the distance to a line segment, the distance to a curved segment is in general difficult. We have to solve degree (2n-1) of the polynomial to find the closest point on a degree n curve, so we have problems of the computational cost and robustness. We propose an effective algorithm for calculating the closest point on a Bezier curve by using the Bezier Clipping method which is an iteration method using only linear equations.
The method proposed here can be applied to wide variety of cases such as pint inversion for curves (finding parameter values at specified points on curves), selecting the closest curve within multiple curve, and the interactive deforming of curves. As for the applications, this paper also proposes an interactive deformation of curves/images and image morphing, and demonstrates the usefulness of the proposed method by using various examples such as a deformed human body modeled by meatballs.
Additional information
pdf,
pdf,
pdf,
"Lighting Design in Interreflective Environments Using Hopfield Neural
Networks"
by Kentaro Takahashi, Kazufumi Kaneda, T. Yamanaka, Hideo Yamashita,
Eihachiro Nakamae, Tomoyuki Nishita
Abstract
This paper proposes a method for calculating luminous intensity distributions of multiple light sources taking into account both direct and interreflected light when a desired luminance distribution in a space is given. This inverse lighting problem is useful for designing rooms and tunnels. In our approach, a luminance distribution is specified instead of an illuminance distribution, beacuse it is closely connected to the appearance of rooms and is used for lighting sesign in a tunnel. To calculate the intensity distributions of light sourcves quickly and robustly, the property of Hopfield neural networks that thier energy converges to a minimum is exploited.
The proposed method should greatly facilitate lighting design when used with rendering techniques such as the radiosity method. Seceral examples invluding lighting design in a tunnel are shown to demonstrate the usefulness of the proposed method.
Keywords and Phrases:
Lighting Design, Luminous Intensity Distribution, Luminance Distribution, Neural Network, Inverse Problem, Interreflection, Radiosity
Additional information
pdf-file (24Mb)
"Photorealistic Visual Simulation of Water Surfaces Taking into account Radiative Transfer"
by Kazufumi Kaneda, G.Yuan, Eihachiro Nakamae, Tomoyuki Nishita
Abstract
Realistic image synthesis, including water surfaces, is very important for the visual environmental assessment of shore regions. Many techniques for rendering water surfaces have been debeloped, but none of them concern themselves with the color of water surfaces under various conditions such as depth adn transparency of water, weather conditions, etc.
A technique for rendering the color of water surfaces realistically is vital for such visual assessment. This paper proposes a method for rendering realistic water surfaces, especially color, taking into account radiative transfer of light in the water. By using the proposed method, realistic images including water surfaces under various water and weathe conditions can be rendered.
Keyword:
Realistic Image Synthesis, Water surfaces, Radiative Transfer, Color of Water, Water Waves
Additional information
pdf-file (4.8Mb)
"Photorealistic Image Synthesis for Outdoor Scenery"
by Kazufumi Kaneda, T. Okamoto, Eihachiro Nakamae, Tomoyuki Nishita
Abstract
We propose a method for displaying photorealistic images of outdoor scenes suitable for displayint simulation results, such as the appearance of a building under verious weathe conditions. This method portrays the changes in perceived color of buildings under various atmospheric conditions by taking into account the spectral distribution of both direct sunlight and sky light, Such views of buildings (including the effects of atmospheric particles like clouds, fog, and light beams) are useful for the design not only of buildings, but also of whole city areas.
Keywords:
Sky light - Specular reflection - Atmospheric scattering model - Fog effect - Spectral distribution
Additional information
pdf-file (14Mb)
"A Scanline Algorithm for Displaying Trimmed Surfaces by using Bezier Clipping"
by Tomoyuki Nishita, Kazufumi Kaneda, Eihachiro Nakamae,
Abstract
Displaying objects with high accuracy is necessary for CAGD(computer-aided geometric design)
and for the synthesis of photo-realistic images. Traditionally, polygonal approximation methods
have been employed to display free-form surfaces. They bring on low accuracy of display
not only in shape, but also in intensity of objects. In this paper, a scanline algorithm
to directly display parametric surface patches, expressed by trimmed Bezier surfaces, without
polygonal approximation is proposed. In the method proposed here, curved surfaces are subdivided into
subpatches with curved edges intersecting with a scanline, and the intersections of every subpatch
and the scanline are calculated. This method is extremely robust for calculating the intersections,
which can be obtained with only a few iterations; the Bezier clipping method is used for the iteration.
Anti-aliased images with shadows and texture mapping are given to show the effectiveness of
the method proposed.
Keywords:
Bezier surfaces , Scanline algorithm , Robustness , High-quality rendering ,
Surface trimming , Silhouette detection , Shadowing
Additional information
"Geometric Hermite Approximation of Surface Patch Intersection Curves"
by Thomas W. Sederberg, Tomoyuki Nishita
Abstract
This paper introduces two new tools for attacking the problem of approximating the curve of two parametric surface patches: the Bezier clipping algorithm for curve/surface intersection, and heometric Hermite approximation of surface/surface intersection curves. The curve/surface intersection algorithm is used to computer a set of endpoint pairs for all components of the intersection curve. Thereafter, a Gk piecewise parametric approximation of the intersection curve is directly computer with no further subdivision or marching.
An error bound is determined directly from the approximation. If the error is too large, each unsatisfactory approximating curve is split in half and new approximations are made directly.
This procedure is O(h1k+2) convergent, which means that each time the members of a Gk sequence of approximating curves are split in half, the new error is 2-(2k+2) of the previous error (in the limit). Thus, doubling the number of approximating curves in a G3 sequence reduces the error typically by two orders of magnitude(1/256).
Additional information
pdf-file (20Mb)
"Curve Intersection using Bezier Clipping"
by Tom W. Sederberg, Tomoyuki Nishita
Abstract
A technique referred to as Bezier clipping is presented.
This technique forms the basis of anm algorithm for computing the points at which two curbes intersect, and computing the points at which two curbes intersect, and also an algorithm for robustly and quickly computing points of tangency between two curbes. Bezier clipping behaves like an intelligent interval Newton method, in which geometric insight is used to identify regions of the parameter domain which exclude the solution set.
Implementation tests suggest that the curve intersection algorithm based on implicitization (though faster than other algorithms) for curves of degree four and less, and is faster than the implicitization algorithm for highter degrees.
bezier
Keywords:
clipping, curve intersection, tangency, focus, polynomial, collinear normal algorithm
Additional information
pdf-file (10.0MB)
pdf,
"A Color Perspective Expression of Illuminance Distribution Illuminated by Multi‐ point Sources"
by Tomoyuki Nishita and Eihachiro Nakamae
Abstract
This paper proposes the following two techniques for lighting design; color-shaded perspectives of three dimensional objects, which are lighted up with several different types of light sources, and isolux diagrams, which are expressed by color belts consisting of different colors depending on illuminance levels, overlapped with the perspectives.
The advanced points of this paper are as follows:
Firstly, displaying perspectives observed from arbitrary viewpoints, designers can easily examine luminous intensity, color, and arrangement of light sources in order to search the optimal lighting effects; that is, the designers can closely watch the whole or local views by moving their viewpoint and center of vision. Secondly, by overlapping an isolux diagram with color belts and the perspectives, the designers can easily grasp the illuminance distribution of every part.
Futhermore, by picking an arbitrary part from the picture, extending it into the whole size, and subdividing each isolux belt, the designers can get more precise data.
Additional information
pdf,
"A Screen Subdivision Method for Half-Tone
Representation of 3-D Objects Using Mini-Computers,"
by Tomoyuki Nishita, Eihachiro Nakamae
Abstract
A half-tone representation of three-dimensional objects is a useful tool for architectural, mechanical, and lighting designs. In many cases, however, a mini-computer is not available for such complex scenes consisting of many objects and/or light sources.
This paper describes a method suitable for a mini-computer on half-tone representation of complex scenes ;a display method by means of subdividing a screen into several parts and processing each part by using only the minimum required date for it, is proposed. The proposed method is very useful not only for saving memory, but also for shortening computation time.
This paper discusses the methods for subdividing a screen and the procedures which extract objects existing in the region of a subdivided screen and objects casting shadows from the outside of that region onto the subdivided screen.
Additional information
pdf-file (4.5Mb)
"An Algorithm for Half-Toned Representation of Three Dimensional Objects"
by Tomoyuki Nishita, Eihachiro Nakamae
Abstract
The probrems concerning hidden line elimination and half toned presentation have to be solved to realistically display three dimensional objects on two dimensional planes. However, these problems are usually neccessaryu considerably complex input data and programs, and a hreat amount of computing time.
Therefore, from the view points of man-machine systems. these problems have not been solved sufficiently.
The authors presented an algorithm for hidden line elimination of polyhedra which were divided into proper convex polyhedra, and showed that the program based on this algorithm has ability of saving computing time. Expanding the algorithm mentioned above, a half toned presentation algorithm for three dimensional objects illuminated with parallel light are proposed, and some examples are expressed to estrablish the usefulness of this algoorithm.
The merits of this algotirhm are as follows; 1) as an object is divided into proper convex polyhedra and they are standardized, the preparation for input data is very easy and the data errors can be reduced compared with traditional methods, 2) the procedures for changing sizes and positions of convex polyhedra are very simple and the some of them can be generated automativcaly, 3) the algorithm of half toned representation is simple, and the computing time is saved by making a goor use of the property of the contour lines of convex polyhedra.
Additional information
"An Algorithm for Hidden Line Elimination of Polyhedra"
by Eihachiro Nakamae, Tomoyuki Nishita
Abstract
Since I. E. Sutheland pointed out the need of eliminating the hidden portions of the three dimensitonal figures on the occasion of displaying on a plane, many papers have been published.
The authors simplified the decision of whether each plane is front or back and of the domains of a convex polyhedron divided an object suitably. Furthermore, the decision for eliminating of hidden line about the innner edges of the independent convex polyhedra for each other is unnecessary.
The merits of this algolithm are as follow; 1) The input data are so simple that its errors are decreased exceedingly. 2) The procedure for hidden line elimination is so simple, because a contour line of each convex polyhedron is convex, too. 3) The design for the structures are very easy, as the position and size of each convex polyhedron is able to modify individually. 4) The figures on the plane of an object are able to be dealed on the same algorithm as the inner edges.
Additional information
"A Simplified Plane-Parallel Scattering Model and its Application to Hair Rendering"
by Mikio Shinya, Mishio Shiraishi,
Kei Iwasaki, Yoshinori Dobashi, Tomoyuki Nishita,
Abstract
Fast computation of multiple light reflections and
scattering among complex objects is very important in photorealistic
rendering. This paper applies the plane-parallel scattering
theory to the rendering of densely distributed objects
such as hairs. We propose a simplified plane-parallel model
that has very simple analytic solutions. This allows us to deal
with multiple scattering phenomena in hair under ambient and
area light illumination. The model was successfully applied to
hair rendering, and experiments demonstrated the efficiency
of the proposed method.
Additional information
"A Method for Creating Mosaic Images Using Voronoi Diagrams"
by Yoshinori Dobashi, T. Haga, H. Johan, Tomoyuki Nishita,
Abstract
This paper proposed a non-photorealistic rendering method that creates an artistic effect called mosaicing. The proposed method converts provided by the user into the mosaic images. Commercial image editing applications also provide a similar funtion. However, these applications often trade results for low-cost computing. It is desirable to create high quality images even if the computational cost is increased.
We present an automatic method for mosaicing images by using Voronoi diagrams. The Voronoi diagrams are optimized so that the error between the original image and the resulting image is as small as possible. Next, the mosaic image is generated by using the sites and edges of the Voronoi diagram.
We use graphics hardware to efficiently generate Voronoi Diagrams. Furthermore, we extend the method to mosaic animations from sequences of images.
"Interpolating 2D Shape Hierarchically"
by Henry Johan, Tomoyuki Nishita,
Abstract
Shape interpolation has been widely used in the field of computer graphics for modeling and for creating visual effects. This paper presents a novel hierarchical method to interpolate between two 2D shapes. A hierarchical shape representation, which is a hierarchy of triangles, is proposed to represent the interior and the detatils of each shape.
By constructing the compatible hierarchical representations of the two shapes, the intermediate shapes are computed by interpolating the corresponding triangles at the lowest level to the highest level of the representations. From experimental results, the proposed method produces interpolation sequences.
"Simulation of Cumuliform Clouds Based on Computational Fluid Dynamics"
by Ryo Miyazaki, Yoshinori Dobashi, Tomoyuki Nishita
Abstract
Simulation of natural phenomena is one of the important research fields in computer graphics. In particular, colouds play an important role in creating images of outdoor scenes. Fluid simulation is effective in creating realistic colouds beacuse colouds are the cisualization of atmospheric fluid.
In this paper, we propose a simulation technique, based on a numerical solution of the partial differential equation of the atmospheric fluid model, for creating animated cumulus and cumulonimbus clouds with features formed by turbulent vortices.
"Adaptive solid texturing for Web3D applications"
by B.-Y. Chen, Tomoyuki Nishita,
Abstract
Solid texturing [Peachy 1985; Perlin 1985] has become a well-known ocmputer hgraphics technology since it was girst presented more than fifteen years ago. Howewver, solid texturing still rematins problems today, because it consumes too much time and has a very high memory requirement. Although some methods have recently been proposed to solve these problems. almost al of them need the support of hardware accelerators.
Hence, these methods could not be applied to all kinds of machines, especially the low-cost ones available over th Internet. Therefore, we present a new method for procedural solid texturing in this paper.
Our approach could almost render an object with procedural solid texturing in real-time using only a software solution. The basic idea of this approach is similar to the cache mechanism usrd for main memory control. Fu's implementation, since this could not receive any benefit from the harware and could be executed on the Internet directly.
<
Additional information
"Interactive Rendering of Shafts of Light Using a Hardware-Accelerated Volume Rendering Technique"
by Y. Dobashi, T. Okita, T. Nishita
Abstract
Recently, graphics hardware has increased in capability and is available even on standard PCs.
These facts have encouraged researchers to debelop hardware-accelerated mehods for rendering realistic images.
One of the important elements in enhancing reality is the effects include light beams from spotlights and shafts of light from the sun's displaying shafts of light at interactive rates by making use of graphics hardware.
"Modeling of Clouds Using a Coupled Map Lattice"
by Yoshinori Dobashi, Ryo Miyazaki, Satoru Yoshida, Tomoyuki Nishita
Abstract
Clouds play an important role in creating images of outdoor scenes. This abstract proposes a new method for modeling various kinds of clouds. The shape of clouds is determined by atmospheric fluid dynamics.
This implies that various types of clouds can be modeled by a physically based simulation of cloud formation. Our method simulates cloud formation by using a method called the coupled map lattice(CML). The proposed mehotd can create various types of clouds and can also realize the animation of these clouds.
Last update: 10 Dec. 2002