STAR Chairs: Dieter Fellner (D), Roberto Scopigno (I)
Day: Wednesday, Sep 4 Time: 10:30-12:00 Room: CS H2
S1: Global Illumination for Interactive Applications and High Quality Animations
Using global illumination algorithms significantly improves the realism of synthetic images. Unfortunately, a vast majority of the existing algorithms were designed for rendering static scenes: when used for dynamic scenes, all computations have to be repeated from scratch even when only minor changes are performed. Clearly, to be of practical use, modern global illumination algorithms have to address explicitly the problem of global illumination computation for changing environments, by taking into account temporal coherence.
In recent years, several new algorithmic solutions have been proposed.
Most of them aim at either shortening production time for high quality
movie sequences, or at providing a fast feedback to the user in interactive
scenarios, in order to improve the usability of lighting simulations. Both
the hardware and algorithms at hand are now mature enough to consider feasible
lighting simulations for animated environments. This state of the art report
is therefore intended to provide a much needed classification and comparison
of the most up-to-date global illumination techniques for animated scenes.
Day: Wednesday, Sep 4 Time: 16:00-17:30 Room: CS H1
S2: Interactive High-Quality Volume Rendering with Flexible Consumer Graphics Hardware
Recently, the classic rendering pipeline in 3D graphics hardware has become flexible by means of introducing programmable geometry engines and rasterization units. This development is mainly driven by the mass market of computer games and entertainment software, whose demand for new special effects and more realistic scenes induced a reconsideration of the static rendering pipeline. Besides the impact on scene complexity in computer games, these advances in flexibility provide an enormous potential for new volume rendering algorithms and allow yet unseen quality in texture-based volume rendering approaches as well as improved performance.
This STAR report covers the latest, advanced algorithms for volume rendering that employ new features and extensions of consumer graphics hardware to achieve both, high quality and interactive frame rates. Beginning with basic texture- based approaches, the algorithms are improved and expanded incrementally, covering pre- and post-classification, illumination, non- polygonal isosurfaces, multi-dimensional transfer functions, pre-integration and volumetric effects.
Especially techniques like pre-integration and multi-dimensional or
high-frequency transfer functions allow new insights into scientific volume
data by allowing enhanced quality and by visualizing hidden features contained
within the the data. Additionally, due to reduced rasterization requirements
and thereby increased performance, better interactivity in scientific volume
visualization and new realistic volumetric effects in entertainment applications
Day: Thursday, Sep 5 Time: 10:30-12:00 Room: CS H1
S3: Visual Data Mining
Visual Data Mining is a novel approach to data mining. It denotes the
combination of traditional data mining techniques and information visualization
methods. The utilization of both the automatic analysis methods and human
perception and understanding promises more efficient data mining techniques.
In this STAR report we provide a systematic overview on the field of Visual
Data Mining. We introduce the common terminology and present an appropriate
framework integrating several relevant models from information visualization.
We discuss the potential and limitations of information visualization and
present a classification and overview on relevant techniques. Finally,
we introduce approaches to integrate information visualization with traditional
mining techniques and point out the benefits and specific challenges of
such an approach.
Day: Thursday, Sep 5 Time: 14:00-15:30 Room: CS H1
S4: Feature Extraction and Visualisation of Flow Fields
We will survey the state of the art in flow visualisation. This is a challenging field, because of the high complexity and dynamics of the flow phenomena, and the huge size of the data sets. We will divide the flow visualisation techniques into three groups: direct or global techniques, geometric techniques, and feature-based techniques. With direct techniques a large part of a flow field is shown in an intuitive way. For geometric techniques, geometric objects such as curves, surfaces or volumes are calculated from the field for visualisation. In feature-based visualisation, important phenomena, such as vortices or shock waves, are extracted from the data, quantitatively described, and visualised. We will survey existing techniques in each category, with an emphasis on the last category.
We will discuss different approaches to feature extraction, and describe
techniques for quantifying phenomena. In time-dependent data sets, correspondence
between persistent features in subsequent time steps has to be established.
Also, important events in the evolution of flow features can be detected.
We will discuss time tracking and event detection techniques in time-dependent
flow data sets.
Day: Friday, Sep 6 Time: 10:30-12:00 Room: CS H3
S5: Tone Reproduction and Physically Based Spectral Rendering
The ultimate aim of realistic graphics is the creation of images that provoke the same responses that a viewer would have to a real scene. This STAR addresses two related key problem areas in this effort which are located at opposite ends of the rendering pipeline, namely the data structures used to describe light during the actual rendering process, and the issue of displaying such radiant intensities in a meaningful way.
The interest in the first of these subproblems stems from the fact that it is common industry practice to use RGB colour values to describe light intensity and surface reflectancy. While viable in the context of methods that do not strive to achieve true realism, this approach has to be replaced by more physically accurate techniques if a prediction of nature is intended.
The second subproblem is that while research into ways of rendering images provides us with better and faster methods, we do not necessarily see their full effect due to limitations of the display hardware. The low dynamic range of a standard computer monitor requires some form of mapping to produce images that are perceptually accurate. Tone reproduction operators attempt to replicate the effect of real-world luminance intensities.
This STAR report will review the work to date on spectral rendering
and tone reproduction techniques. It will include an investigation into
the need for spectral imagery synthesis methods and accurate tone reproduction,
and a discussion of major approaches to physically correct rendering and
key tone mapping algorithms. The future of both spectral rendering and
tone reproduction techniques will be considered, together with the implications
of advances in display hardware.