Catálogo de publicaciones - libros

We present a system using semi-autonomous agents to help artists express ideas. Agents control their representation on a canvas via interactions in agent space. They are given rules, which include the ability to form, join, leave and dissolve coalitions. While individually, agents constitute only small parts of the composition, together they form more and more complex parts until the full picture emerges. The artist can take direct control of an agent or coalition if the result is unsatisfactory. We have implemented the Surreal engine that realises these ideas and we present a case study of its usage in producing Mondriaan-like paintings.

Co-referential relations between textual and visual elements in illustrations can be encoded efficiently through textual labels. The labels support students to learn unknown terms and focus their attention on important aspects of the illustration; while a functional and aesthetic label layout aims at guaranteeing the readability of text strokes as well as preventing the referential mismatches. By analyzing a corpus of complex label layouts in hand-drawn illustrations, a classification of label layout styles and several metrics for functional requirements and aesthetic attributes were extracted. As the choice of a specific layout style seems largely determined by individual preferences, a real-time layout algorithm for internal and external labels balances conflicting user-specific requirements, functional and aesthetic attributes.

This paper presents a smart algorithm for labeling column charts and their derivatives. To efficiently solve the problem, we separate it into two sub-problems. We first present a geometric algorithm to solve the problem of finding a good labeling for the labels of a single column, given that some other columns have already been labeled. We then present a strategy for finding a good order in which columns should be labeled, which repeatedly uses the first algorithm for some limited look-ahead. The presented algorithm is being used in a commercial product to label charts, and has shown in practice to produce satisfactory results.

Due to the progress in computer graphics hardware high resolution 3d models may be explored at interactive frame rates, and facilities to explore them are a part of modern radiological workstations and therapy planning systems. Despite their advantages, 3d visualizations are only employed by a minority of potential users and even these employ 3d visualizations for a few selected tasks only. We hypothesize that this results from a lack of intuitive interaction techniques for 3d rotation. In this paper, we compare existing techniques with respect to design principles derived by clinical applications and present results of an empirical study. These results are relevant beyond clinical applications and strongly suggest that the presented design principles are crucial for comfortable and predictable interaction techniques for 3d rotation.

Parametric design systems model a design as a constrained collection of schemata. Designers work in such systems at two levels: definition of schemata and constraints; and search within a schema collection for meaningful instances. Propagation-based systems yield efficient algorithms that are complete within their domain, require explicit specification of a directed acyclic constraint graph and allow relatively simple debugging strategies based on antecedents and consequents. The requirement to order constraints appears to be useful in expressing specific designer intentions and in disambiguating interaction. A key feature of such systems in practice appears to be a need for multiple views onto the constraint model and simultaneous interaction across views. We describe one multiple-view structure, its development and refinement through a large group of architecture practitioners and its realization in the system Generative Components.

Cross-sectioning is a popular method for visualizing the complicated inner structures of three-dimensional volume datasets. However, the process is usually manual, meaning that a user must manually specify the cross-section’s location using a repeated trial-and-error process. To find the best cross-sections, this method requires that a user is knowledgeable and experienced. This paper proposes a method for automatically generating characteristic cross-sections from a given volume dataset. The application of a volume skeleton tree (VST), which is a graph that delineates the topological structure of a three-dimensional volume, facilitates the automated generation of cross-sections giving good representations of the topological characteristics of a dataset. The feasibility of the proposed method is demonstrated using several examples.

Large horizontal displays provide new opportunities to support individual and collaborative activities such as creativity and organizational tasks. We present Interface Currents, a fluid interaction technique designed to support face-to-face collaboration by improving access to and sharing of workspace items. Interface Currents are flexible containers that provide a controllable flow of interface items that support creativity during collaborative tasks and enable intuitive organization and sharing of digital information around horizontal displays.

Indirect and unintuitive commands like mode-switching are difficult operations for many users. In this paper, we describe a new simple interface for a 2D drawing system to lessen user’s confusion. Firstly, we propose a direct manipulation interface based on a set of geometric primitives and their geometric relationships. Secondly, we present a seamless drawing interface that users do not need to switch modes by clicking the toolbar. Through user tests using our prototypes and a standard commercial application, we confirmed that our system can complete certain drawing tasks with shorter completion time and less number of commands.

We describe our current work on volume visualization techniques developed with the aim to enhance interactive exploration of medical datasets. Volumetric lenses can be applied to a volume dataset to interactively focus regions of interest within these datasets. During rendering the parts of a volume dataset intersecting the lens, which is defined by a convex 3D shape, are rendered using a different visual appearance. The lenses proposed allow to apply non-photorealistic rendering techniques interactively to aid comprehension for medical diagnosis.

3d visualisation of real spaces in the Internet is nowadays getting ready for all day usage. Because of higher complexity of 3d- compared to 2d visualisation, it is mostly used for high-end visualisations. The strategy of integrating them into existing spatial data infrastructures and making reuse of existing resources that are so far mostly used for 2d visualisation, created the opportunity to build 3d web-applications as an add-on. The recurring problems of web-based 3d visualisation raised the question, how special modules for 3d visualisation could also be made reuseable. We took a closer look especially on the so-called mapping-level, which is essential for the type of visualisation, interactivity and some special analysis-tasks.