Contributions published at Visualization and MultiMedia Lab (Renato Pajarola)

A recent trend in astronomy is the hunt for a second life sustaining planet in the cosmos. This research generates a lot of data of celestial bodies, so-called extrasolar planets or exoplanets, which is visualized here. Concepts for the adequate representation of the discovered objects are developed and implemented as an interactive computer application called ExoViewer. Apart from planetary bodies, the program illustrates Asteroids and Comets present in our Solar System. The challenge of rendering this huge amount of data in an appropriate manner is faced as a second part of the work. Results, which are frames rendered using described techniques and methods, show that visualizing uncertain or missing data is a big issue and the focus of a user can quickly get distracted by displaying too many things. Additionally, measurements of the performance of the application illustrate that calculating positions of bodies in space are costly operations and culling strategies are crucial to obtain a smooth running program.

null

null

The implementation of traffic data is an important and vital visualization research field. With an understandable visualization a user is able to acquire the information in a dataset faster and more accurately. In this report, the design, planning and evaluation of a web-based visualization application is provided. The data-set used is the real-time API from SBB and the animations are written in Javascript with WebGl. A server API was created that works together with a client side implementation of the visualization. The core of the report is the technical setup and the implementation. There the technical solution, the design cycles, the server and client implementation are included. Then the experimental results demonstrate the state of the application and the different possibilities that exist for a user to interact with the visualization. What can a user query and how can the visualization provide a meaningful output that can be analyzed. The user has to be able to estimate the arriving of a train at a station and estimate where he can travel in a certain timeframe. This project developed a web based visualization with a realtime API on a map visualization on which a user can query a departing station and receive multiple connection possibilities. This was done using different visualization modes on top of a map API.

Reconstructing the as-built architectural shape of building interiors has emerged in recent years as an important and challenging research problem. An effective approach must be able to faithfully capture the architectural structures and separate permanent components from clutter (e.g. furniture), while at the same time dealing with defects in the input data. For many applications, higher-level information on the environment is also required, in particular the shape of individual rooms. To solve this ill-posed problem, state-of-the-art methods assume constrained input environments with a 2.5D or, more restrictively, a Manhattan-world structure, which significantly restricts their applicability in real-world settings. We presents a novel pipeline that lifts these assumptions and allows to reconstruct general 3D interior architectures, significantly increasing the range of real-world architectural shapes that can be reconstructed and labeled by any semi-automatic method to date. Our method finds candidate permanent components by reasoning on a graph-based representation of the scene, then uses them to build a 3D linear cell complex that is partitioned into separate rooms through a multi-label energy minimization formulation. We demonstrate the effectiveness of our method by applying it to a variety of real-world and synthetic datasets and by comparing it against the more specialized state-of-the-art approaches.

null

Tensor decompositions, in particular the Tucker model, are a powerful family of techniques for dimensionality reduction and are being increasingly used for compactly encoding large multidimensional arrays, images and other visual data sets. In interactive applications, volume data often needs to be decompressed and manipulated dynamically; when designing data reduction and reconstruction methods, several parameters must be taken into account, such as the achievable compression ratio, approximation error and reconstruction speed. Weighing these variables in an effective way is challenging, and here we present two main contributions to solve this issue for Tucker tensor decompositions. First, we provide algorithms to efficiently compute, store and retrieve good choices of tensor rank selection and decompression parameters in order to optimize memory usage, approximation quality and computational costs. Second, we propose a Tucker compression alternative based on coefficient thresholding and zigzag traversal, followed by logarithmic quantization on both the transformed tensor core and its factor matrices. In terms of approximation accuracy, this approach is theoretically and empirically better than the commonly used tensor rank truncation method.

Dense line graphs and polyline maps are challenging for interactive visualization in geographic information systems (GIS). Bundling techniques are a common approach to reduce clutter and have successfully been demonstrated for the display of complex planar graphs. Previous techniques typically employed some form of attraction or repulsion forces to bundle edges in two dimensions, and while in principle extensible to 3D they do not directly support hard intersection constraints in a 3D environment. In geographic visualization systems, e.g. such as interactive virtual globes or 3D GIS viewers, it is often necessary to take the 3D environment into account and to: (1) bundle lines and paths in 3D, (2) constrain path bundles to follow some reference network vector map, as well as (3) avoid intersections with the digital elevation model (DEM). In this paper we introduce a novel method which uses geographic vector map reference information to route, visualize and simplify path bundles along their network paths in a constrained 3D environment using adaptive B-splines. Moreover, we describe an efficient rendering architecture to flexibly display bundled paths within a 3D rendering pipeline at varying level of detail (LOD).

Terrain rendering is widely used in industry and research. GIS software packages as well as navigation systems make use of terrain rendering to visualize terrain information. Recent trends in research show that scientific terrain visualization is shifting more and more to an interactive analysis tool. This allows domain specific users to perform visual analysis tasks within an interactive visual environment. Visual analysis tools are software package acting as a toolbox and providing functionality to support the work of domain specific users such as data exploration, data analysis and data presentation. Such software packages still suffer from limitations such as restricted or imprecise data and problems with large data handling. These challenges will also be at the core of research in scientific terrain visualization in the near future. In this paper we describe some open challenges for scientific terrain visualization in the acquisition, processing and rendering of terrain related geospatial information as well as new methods which could be used to address these challenges.

This thesis introduces the Atmosphere Viewer which lets the spectator simulate atmospheres on Earth and other planets in our solar system based on their real properties to our knowledge - presenting an interdisciplinary project with the combination of Informatics, Mathematics, Physics, Chemistry, Geometry and Optics.

Using many lights in real-time applications has been an important goal for many years. The games industry, in particular, has strived to increase the number of lights to provide enhanced visual quality and realism. Today, high-end games often make use of hundreds of lights in each frame, and the number of lights is likely to expand in the future. The ability to efficiently manage and shade large numbers of lights presents many possibilities, apart from simply allowing light to be cast from many dynamic objects. It can also support visualizing global illumination solutions, or enable detailed artistic light direction. Efficient real-time shading with many lights helps solve several problems facing the development of next-generation high-end games. To achieve the level of performance needed to make this possible, light management and shading have evolved dramatically in recent years. Both industry and academia have invested significant effort in pursuit of this goal, which has resulted in a large number of new and sometimes competing techniques. This course presents an in-depth exploration of many-light management, starting with background and progressing to state-of-the-art research, including recent results on supporting shadows. The course combines production experience from game developers with the latest research into efficient many-light algorithms for both desktop and mobile hardware.

In this thesis a method is proposed that performs regional color transfer on 2D laser-scanned range images of unlike scenes by matching cells of manually defined reference regions to cells of manually defined target regions. The matching between cells is performed by comparison measures based on feature detection algorithms like Histogram Comparison, Normalized Cross-Correlation and Histograms of oriented Gradients. The color transfer is performed via statistical per-channel color transfer. The proposed approach is partially automatic by requiring the user to select target and reference areas and setting the weights for the different comparison methods. The approach as well as the feature detection algorithms are evaluated and their suitability on the subject of laser range scanners is analyzed.

Since nowadays computer generated characters are so common in a variety of different fields, it is highly important to animate their facial expressions in a believable way. Creating them artificially, using keyframe animation, is a very hard and tedious process which often lacks of liveliness; this is why facial performance capture is used more and more. The goal of this thesis is to develop a system that enables the user to capture a facial performance and let it drive a non-human character in an inexpensive and easy way. Therefore we developed a plug-in for Autodesk Maya that is capable of capturing the actor's emotions from video files, using marker tracking, and after a configuration step, directly apply them as blend shapes onto the target character. The system delivers convincing results with the focus on non-human characters that have a completely different facial anatomy than humans.

We propose a new method to enhance the depth images from RGB-D sensors, such as Kinects, by filling the missing/invalid values which are reported by those sensors at certain pixels. We introduce a robust 1D least median of squares (1D LMedS) approach to accurately estimate the depth values of those invalid pixels. We use this approach for efficient traversal of each pixel’s depth values over a sequence of frames and look for invalid depth values (considered as outliers), and finally, replace those values with stable valid depth values. Our approach solves the unstable nature of depth values in captured scenes that is perceived as flickering. Experimental results show good improvement both for static and moving parts of a scene.

In this paper, a new method to extract wall openings (windows and doors) in interior scenes from point clouds under cluttered and occluded environments is presented. For each wall surface or a room represented by a bounding polyhedron and its 3D scan points, our method constructs a planar cell complex representation, which is used for the wall features segmentation using a graph-cut method. We evaluate our results of the proposed approach on real-world 3D scans of indoor environments and demonstrate its validity.

We propose a new technique for in-core and out-of-core GPU ray tracing using a generalization of hierarchical occlusion culling in the style of the CHC++ method. Our method exploits the rasterization pipeline and hardware occlusion queries in order to create coherent batches of work for localized shader-based ray tracing kernels. By combining hierarchies in both ray space and object space, the method is able to share intermediate traversal results among multiple rays. We exploit temporal coherence among similar ray sets between frames and also within the given frame. A suitable management of the current visibility state makes it possible to benefit from occlusion culling for less coherent ray types like diffuse reflections. Since large scenes are still a challenge for modern GPU ray tracers, our method is most useful for scenes with medium to high complexity, especially since our method inherently supports ray tracing highly complex scenes that do not fit in GPU memory. For in-core scenes our method is comparable to CUDA ray tracing and performs up to 5.94× better than pure shader-based ray tracing.

This thesis presents the Solar System Viewer, a data visualization application visualizing our solar system in three dimensional space. The application allows the user to observe the various celestial bodies like planets or moons and their orbits. While doing so, the user can move through both space and time as they wish. Using the parallel rendering framework Equalizer, the rendering of the Solar System Viewer gets parallelised in order to allow the application to be displayed across multiple displays on a display wall. After a look at the fundamentals of parallel rendering, the necessary astronomical concepts are explained. Then the user interface, the data model and the implementation of the Solar System Viewer are described. In the end the thesis takes a look at existing applications with a functionality similar to that of the Solar System Viewer.

null

We propose a new method to fill missing or invalid values in depth images generated from the Kinect depth sensor. To fill the missing depth values, we use a robust least median of squares (LMedS) approach. We apply our method for telepresence environments, where Kinects are used very often for reconstructing the captured scene in 3D. We introduce a modified 1D LMedS approach for efficient traversal of consecutive image frames. Our approach solves the unstable nature of depth values in static scenes that is perceived as flickering. We obtain very good result both for static and moving objects inside a scene.

Recently, several algorithms have been introduced that enable real-time performance for many lights in applications such as games. In this paper, we explore the use of hardwaresupported virtual cube-map shadows to efficiently implement high-quality shadows from hundreds of light sources in real time and within a bounded memory footprint. In addition, we explore the utility of ray tracing for shadows from many lights and present a hybrid algorithm combining ray tracing with cube maps to exploit their respective strengths. Our solution supports real-time performance with hundreds of lights in fully dynamic high-detail scenes.