/ Carl W. Vilbrandt / Associate Professor
/ Sergei I. Vyatkin / Visiting Associate Professor
The Computer Arts Lab is dedicated to developing computer mastery. Computer arts are analogous to industrial arts or martial arts. Thus, skill and craft in the computer arts are acquired in the practice of using and applying computers in new, creative and challenging ways requiring close and insightful observation.
Computer arts are developed through exploration, observation, engineering and invention that invokes emotion and thought, and augments the physical and spiritual nature of things. To be a master of computer arts, one must be a master of his or her computer hardware and software tools, and that is most effectively done by modification or creation of hardware and software tools. The Computer Arts Lab develops the students' computer skills by using the latest developments in computer science, whenever possible, in collaboration with other University of Aizu Computer Labs. Computer Arts Lab students are challenged to be a technical team; they are asked to select or design new computer applications and take responsibility for the incremental implementation of their selection or design.
The wide angle view of the Computer Arts Lab is very focused on its primary target of personal achievement of a high level of computer fluency for its students. The Computer Arts Lab program includes the free and open design and implementation of computer hardware and software under the GGPL agreement.
Our program supports the development of the Fukushima hardware and software initiatives, the HyperFunction Consortium for FRep, the Aizu History Project, and the DALI conference for open and free exchange of information. In turn, these associations encourage a stimulating environment of international research for student projects that meet and support the general objectives above. Student projects are divided into simple tasks and micro managed and implemented by University of Aizu faculty, students, local and international businesses, and governmental organizations. Each project will seek separate funding and support from local, national and international sources.
Refereed Journal Papers
Since the human eye is spherical, the best visual systems for computer simulations might be dome planetariums or spherical projections on the viewer's eyes rather than the usual flat screen displays. Spherical projection on a twenty foot diameter sphere, the limits of the acuity of the eye, or separate spherical projections on each of the viewers eyes are required for dramatically improving computer simulations for immersive environments. Real time spherical projected images need to be free of visual distortion. The flat screen linear algorithms are not sufficient to meet real time spherical projections' distortion free requirements. The distortion and visual artifacts created by the projection lens, the non planar dome or the variation in the eyes of the viewer, and the projector's electronics are quite complex and require the use of non-linear algorithms in resolving the spherical mapping. Our approach is oriented on tasks that require extreme non-linear computational capabilities. However, the proposed approach is promising and should be considered for future development that could dramatically improve visual technology for immersive environments.
Using the example of the no longer extant Golden Hall at Enichiji in the Aizu region of Japan, we demonstrate the construction of architectural models based on archaeological evidence. We discuss the process of decision-making in cases in which evidence is fragmentary or conflicting. We also demonstrate beginning work on converting such models to formats suitable for user-directed virtual reality walkthroughs, in particular through the open source gaming engine Quake 2.
Convolution surfaces enable the user to model complex free-form shapes. Due to analytical solutions for some kernel functions and skeletal elements, it is possible to model convolution surfaces interactively. An extendable user interface allows the user to design models using different types of convolution surfaces. New primitives can be easily bound to the modeller using the proposed blending technique. Models generated in the HyperFun language can be exchanged between modelling tools on several platforms.
Interactive design of a n-ary tree data structure for FRep (functionally represented) geometric models is discussed. The interactive Construction Tree tool which is a part of an FRep modeler is designed to construct a tree structure and convert it to a HyperFun language description. With the developed tool, a model in HyperFun can be also read with a corresponding graphical tree displayed. The user can interact with the 3D geometric model through the graphical tree. Using the interactive design of a construction tree, the user will be able to create HyperFun models more easily than before. The geometric data structures will be incorporated into a design database for intended use on clusters of computer servers allowing low end clients access to advanced geometric modeling over the Internet.
Many simulator applications, visual systems for virtual environments require the generated image to be free of distortion regardless of where in some allowable volume the observer's eyepoint lies. There is a solution of the given problem very important for image transfer directly into the eye of the observer by laser. In the future, personal laser projection devices can change a personal computer display. Since the distortion mapping is in general different for each eyepoint, some means of locating the eyepoint is necessary. A head-tracking device fitted to a helmet is the usual solution. A more difficult problem lies in determining what the distortion mapping looks like from each viewpoint and inverse mapping. Our approach is oriented on tasks that require extreme computational capabilities. It is a promising one for future development taking into consideration dramatically improving technology.
The HyperFun Project is devoted to developing an open system architecture for functionally based (implicit or more generally FRep) shape modeling and its applications. The software tools are built around the shape models written in a high-level programming language called HyperFun. A model in HyperFun can serve as a protocol for exchanging FRep models between users, modeling systems, or networked computers. HyperFun models can be collected in application-specific libraries. We describe the basic set of system components: an interpreter for parsing and function evaluation; FRep system libraries; a modeler with an extendable graphical user interface; a multidimensional modeler with a symbolic user interface providing means for interpreting multidimensional coordinates and constructing scenes; applications for visualization (polygonization, VRML generation, ray-tracing), animation, voxelization and others; a collaborative Internet-based modeler including a HyperFun-to-Java translator and advanced interactive techniques based on the empirical modeling paradigm. These components are intended to be public domain to stimulate collaborative development efforts (see www.hyperfun.org).
The open source movement seems to defy logic at times. We explore the age of computing in which people build high-quality software and open hardware design for the love of it and then give it away. Rather than focus on details like licensing and software toolkits, we try to deliver a basic framework or rules of engagement for life in the open source computing environment. We talk about: why the open source movement can't be ignored; why it's critical to understand the basic principles of geek culture; why lying is unforgivable in the open source community; why the open source community is self-correcting in more than just code; why the community tolerates some degree of rudeness; why publicly admitting errors is critical to your success; why using some standard types of business communication can leave you stunned and shunned.
The ACM1 Exhibition had 25,000 visitors; 13 researchers from 5 counteries presented the HyperFun Project, a joint exhibit between the University of Aizu and Hosei University. The HyperFun language was introduced for teaching and practical use of FRep modeling. It is a minimalist programming language supporting all notions of FRep. The application software deals with HyperFun models through the built-in language interpreter or using HyperFun-to-C/C++/Java compiler and utilities of the HyperFun API. Software tools are being developed in an open source project manner by the international team of developers. Some of them are currently available for free download at http://www.hyperfun.org/: HyperFun Polygonizer for the surface mesh generation with VRML output and HyperFun plug-in to POVRay, which makes it possible to generate high quality photorealistic images on an ordinary PC. These and other experimental tools were demonstrated at the exhibition: interactive modelers of convolution surfaces and 4D volume splines, graphical user interface for FRep constructive tree, and real-time fly through a volumetric object. Further development includes creation of virtual reality and haptic interfaces, special hardware design for visualization of hybrid FRep/voxel models, research on genetic, physics based and finite-element methods for advanced CAD applications.