Animation and Performance Capture using Digitalized Models
(de Aguiar, Edilson, 2008-12-22) de Aguiar, Edilson
The realistic generation of virtual doubles of real-world actors has been the focus of computer graphics research for many years. However, some problems still remain unsolved: it is still time-consuming to generate character animations using the traditional skeleton-based pipeline, passive performance capture of human actors wearing arbitrary everyday apparel is still challenging, and until now, there is only a limited amount of techniques for processing and modifying mesh animations, in contrast to the huge amount of skeleton-based techniques.In this thesis, we propose algorithmic solutions to each of these problems. First, two efficient mesh-based alternatives to simplify the overall character animation process are proposed. Although abandoning the concept of a kinematic skeleton, both techniques can be directly integrated in the traditional pipeline, generating animations with realistic body deformations. Thereafter, three passive performance capture methods are presented which employ a deformable model as underlying scene representation. The techniques are able to jointly reconstruct spatio-temporally coherent time-varying geometry, motion, and textural surface appearance of subjects wearing loose and everyday apparel. Moreover, the acquired high-quality reconstructions enable us to render realistic 3D Videos. At the end, two novel algorithms for processing mesh animations are described. The first one enables the fully-automatic conversion of a mesh animation into a skeleton-based animation and the second one automatically converts a mesh animation into an animation collage, a new artistic style for rendering animations.The methods described in the thesis can be regarded as solutions to specific problems or important building blocks for a larger application. As a whole, they form a powerful system to accurately capture, manipulate and realistically render realworld human performances, exceeding the capabilities of many related capture techniques. By this means, we are able to correctly capture the motion, the timevarying details and the texture information of a real human performing, and transform it into a fully-rigged character animation, that can be directly used by an animator, or use it to realistically display the actor from arbitrary viewpoints.
Efficient Shadow Map Filtering
(Annen, Thomas, 2008-12-12) Annen, Thomas
Shadows provide the human visual system with important cues to sense spatial relationshipsin the environment we live in. As such they are an indispensable partof realistic computer-generated imagery. Unfortunately, visibility determinationis computationally expensive. Image-based simplifications to the problem suchas Shadow Maps perform well with increased scene complexity but produce artifactsboth in the spatial and temporal domain because they lack efficient filteringsupport.This dissertation presents novel real-time shadow algorithms to enable efficientfiltering of Shadow Maps in order to increase the image quality and overallcoherence characteristics. This is achieved by expressing the shadow test as asum of products where the parameters of the shadow test are separated from eachother. Ordinary Shadow Maps are then subject to a transformation into new socalled basis-images which can, as opposed to Shadow Maps, be linearly filtered.The convolved basis images are equivalent to a pre-filtered shadow test and usedto reconstruct anti-aliased as well as physically plausible all-frequency shadows.
Learning-Based Facial Animation
(Bargmann, Robert, 2008-11-28) Bargmann, Robert
This thesis proposes a novel approach for automated 3D speechanimation from audio. An end-to-end system is presented whichundergoes three principal phases. In the acquisition phase, dynamicarticulation motions are recorded and amended. The learning phasestudies the correlation of these motions in their phonetic context inorder to understand the visual nature of speech. Finally, for thesynthesis phase, an algorithm is proposed that carries as much of thenatural behavior as possible from the acquired data to the finalanimation.The selection of motion segments for the synthesis of animationsrelies on a novel similarity measure, based on a Locally LinearEmbedding representation of visemes, which closely relates to visemecategories defined in articulatory phonetics literature. This measureoffers a relaxed selection of visemes, without reducing the quality ofthe animation.Along with a general hierarchical substitution procedure which candirectly be reused in other speech animation systems, our algorithmperforms optimum segment concatenation in order to create newutterances with natural coarticulation effects.
Interactive Illustrative Volume Visualization
(Bruckner, March 2008) Bruckner, Stefan
Illustrationen spielen eine wichtige Rolle in der Kommunikationkomplexer Sachverhalte. Ihre Herstellung ist jedoch schwierigund teuer. Dreidimensionale bildgebende Verfahren habensich in den letzten Jahren als unverzichtbares Werkzeug in Disziplinenwie der medizinischen Diagnose und Behandlungsplanung,im technischen Bereich (z.B. Materialprüfung), der Biologie,und der Archäologie etabliert. Modalitäten wie Röntgencomputertomographie(CT) oder Magnetresonanztomographie (MRT) generierentäglich hochauflösende volumetrische Scans. Es leuchtetnicht ein, dass trotz dieses Datenreichtums die Produktion einerIllustration noch immer ein aufwendiger und zum Großteil manuellerProzess ist.Diese Dissertation beschäftigt sich mit der computerunterstützten Erstellung von Illustrationen direkt auf Basis solcher Volumendaten.Zu diesem Zweck wird das Konzept eines direktenVolumenillustrationssystems eingeführt. Dieses System erlaubtdie Gestaltung einer Illustration direkt anhand von gemessenenDaten, ohne dass ein zusätzlicher Modellierungsschritt notwendigwäre. Abstraktion, ein wichtiger Bestandteil traditioneller Illustrationen,wird verwendet um visuelle Überladung zu vermeiden,wichtige Strukturen hervorzuheben und versteckte Details sichtbarzu machen. Abstraktionstechniken beschäftigen sich einerseitsmit der Erscheinung von Objekten und erlauben die flexiblekünstlerische Schattierung von Strukturen in volumetrischenDatensätzen. Andererseits kontrollieren diese Techniken welcheObjekte sichtbar sein sollen. Neue Methoden zur Generierung vonTransparenz- und Explosionsdarstellungen werden hierfür vorgestellt.Die präsentierten Visualisierungstechniken verwendendie Fähigkeiten moderner Graphikhardware um eine interaktiveDarstellung zu ermöglichen.Das resultierende System erlaubt die Erstellung von expressivenIllustrationen direkt anhand von volumetrischen Daten undhat eine Vielzahl von Anwendungen wie etwa die medizinischeAusbildung, Patientenaufklärung und wissenschaftliche Kommunikation. - Illustrations are essential for the effective communication ofcomplex subjects. Their production, however, is a difficult andexpensive task. In recent years, three-dimensional imaging hasbecome a vital tool not only in medical diagnosis and treatmentplanning, but also in many technical disciplines (e.g., materialinspection), biology, and archeology. Modalities such as X-RayComputed Tomography (CT) and Magnetic Resonance Imaging(MRI) produce high-resolution volumetric scans on a daily basis.It seems counter-intuitive that even though such a wealth of datais available, the production of an illustration should still require amainly manual and time-consuming process.This thesis is devoted to the computer-assisted generation ofillustrations directly from volumetric data using advanced visualizationtechniques. The concept of a direct volume illustrationsystem is introduced for this purpose. Instead of requiring anadditional modeling step, this system allows the designer of anillustration to work directly on the measured data. Abstraction,a key component of traditional illustrations, is used in order toreduce visual clutter, emphasize important structures, and revealhidden detail. Low-level abstraction techniques are concernedwith the appearance of objects and allow flexible artistic shadingof structures in volumetric data sets. High-level abstraction techniquescontrol which objects are visible. For this purpose, novelmethods for the generation of ghosted and exploded views areintroduced.The visualization techniques presented in this thesis employthe features of current graphics hardware to achieve interactiveperformance. The resulting system allows the generation of expressiveillustrations directly from volumetric data with applications inmedical training, patient education, and scientific communication.
HLOD Refinement Driven by Hardware Occlusion Queries
(Charalambos, Dec 2007) Charalambos, Jean Pierre
Con objeto de realizar una eficiente visualización interactiva de modelos geométricoscomplejos, que pueden llegar a comprender varios millones de polígonos, es clave reducirsustancialmente la cantidad de datos procesados. Los métodos LOD (''level-of-detail'')permiten efectuar una agresiva reducción de los datos enviados a la GPU, a expensasde sacrificar algo de calidad visual. Particularmente, los métodos HLOD (''hierarchicallevel-of-detail''), en los que se precomputan LODs en los distintos niveles de una jerarquía espacial dada, han demostrado ser el más eficiente enfoque para la visualizacióninteractiva de este tipo de modelos. Además de soportar en forma directa algoritmos''out-of-core'' (en los que durante tiempo de ejecución los datos deben ser constantementeleídos desde la memoria secundaria del sistema), los métodos HLODs permiten efectuaruna carga óptima de los datos entre la CPU y la GPU.Obedeciendo al mismo objeto, un enfoque ortogonal al anterior es ''occlusion culling''(descarte por oclusión). Respecto a un punto de vista dado se busca descartar de modoeficiente las partes invisibles de la escena y en visualizar solo sus partes visibles. Losmétodos más recientes pertenecientes a esta categoría emplean HOQs (''hardware occlusionqueries'').Los efectos relativos a HLODs y occlussion culling pueden combinarse de modo efectivo.Primero, es posible descartar aquellos nodos de la jerarquía que resulten invisibles.Segundo, para los nodos visibles, es posible emplear los resultados de los HOQs comoparte integral de la condición de refinamiento de la jerarquía: de acuerdo con el grado devisibilidad de un nodo dado y teniendo en cuenta un fenómeno de la percepción denominado''visual masking'', es factible determinar cuando no habría una ganancia apreciableen la apariencia final de la imagen obtenida si el nodo fuera refinado ulteriormente. Eneste caso, HOQs permiten reducir aún más agresivamente el total de primitivas visualizadas.Sin embargo, debido a la latencia presente entre el momento de iniciar el HOQy la disponibilidad de su resultado, el uso directo de HOQs en las condiciones de re-finamiento resultaría siendo una fuente de estancamiento de la CPU, lo que a su vezredundaría en un apreciable desaprovechamiento de la GPU.En esta tesis presentamos una novedosa métrica que emplea información de visibilidad(determinada a partir de HOQs) como parte integral de la condición de refinamiento deun modelo HLOD (de nuestro conocimiento, el primer enfoque en este contexto con estameta en mente). También contribuimos con un novedoso algoritmo para atravesar lajerarquía del HLOD que permite sacar el máximo provecho de esta métrica. A partir deuna rutina básica de predicción de la condición de refinamiento del HLOD, el algoritmominimiza el estancamiento de la CPU y permite obtener así un mejor aprovechamientode la GPU.Las principales propiedades expuestas en nuestro enfoque combinado son: 1. Mejorrendimiento con la misma calidad visual: mediante nuestro sistema es posible visualizarun menor número de primitivas (no por ello nuestra técnica de occlusion culling deja deser conservativa) con una pérdida mínima en la calidad visual del modelo; 2. Generalidad:nuestra métrica soporta cualquier tipo de HLOD; 3. Uso integral de los resultadosobtenidos en HOQs: nuestra métrica aprovecha de modo completo la informaciónobtenida mediante HOQs; 4. Aprovechamiento integral de la coherencia espacio-temporalinherente a las representaciones jerárquicas; y, 5. Implementación directa. - In order to achieve interactive rendering of complex models comprising several millionsof polygons, the amount of processed data has to be substantially reduced. Level-ofdetail(LOD) methods allow the amount of data sent to the GPU to be aggressivelyreduced at the expense of sacrificing image quality. Hierarchical level-of-detail (HLOD)methods have proved particularly capable of interactive visualisation of huge data setsby precomputing levels-of-detail at different levels of a spatial hierarchy. HLODs supportout-of-core algorithms in a straightforward way and allow an optimal balance betweenCPU and GPU load during rendering.Occlusion culling represents an orthogonal approach for reducing the amount of renderedprimitives. Occlusion culling methods aim to quickly cull the invisible part of themodel and render only its visible part. Most recent methods use hardware occlusionqueries (HOQs) to achieve this task.The effects of HLODs and occlusion culling can be successfully combined. Firstly,nodes which are completely invisible can be culled. Secondly, HOQ results can be usedfor visible nodes when refining an HLOD model; according to the degree of visibility of anode and the visual masking perceptual phenomenon, then it could be determined thatthere would be no gain in the final appearance of the image obtained if the node werefurther refined. In the latter case, HOQs allow more aggressive culling of the HLODhierarchy, further reducing the amount of rendered primitives. However, due to thelatency between issuing an HOQ and the availability of its result, the direct use of HOQsfor refinement criteria cause CPU stalls and GPU starvation.This thesis introduces a novel error metric, taking visibility information (gathered fromHOQs) as an integral part of refining an HLOD model, this being the first approachwithin this context to the best of our knowledge. A novel traversal algorithm for HLODrefinement is also presented for taking full advantage of the introduced HOQ-based errormetric. The algorithm minimises CPU stalls and GPU starvation by predicting HLODrefinement conditions using spatio-temporal coherence of visibility.Some properties of the combined approach presented here involve improved performancehaving the same visual quality (whilst our occlusion culling technique still remainedconservative). Our error metric supports both polygon-based and point-basedHLODs, ensuring full use of HOQ results (our error metrics take full advantage of theinformation gathered in HOQs). Our traversal algorithm makes full use of the spatialand temporal coherency inherent in hierarchical representations. Our approach can bestraightforwardly implemented.
A Computational Musco-Skeletal Model for Animating Virtual Faces
(Fratarcangeli, 30/09/2009) Fratarcangeli, Marco
Automatic synthesis of facial animation in Computer Graphics is a challenging task and althoughthe problem is three decades old by now, there is still not a unified method to solveit. This is mainly due to the complex mathematical model required to reproduce the visualmeanings of facial expressions coupled with the computational speed needed to run interactiveapplications.In this thesis, there are two different proposed methods to address the problem of theanimation of 3D realistic faces at interactive rate.The first method is an integrated physically-based method which mimics the facial movementsby reproducing the anatomical structure of a human head and the interaction amongthe bony structure, the facial muscles and the skin. Differently from previously proposedapproaches in the literature, the muscles are organized in a layered, interweaving structurelaying on the skull; their shape can be affected both by the simulation of active contractionand by the motion of the underlying anatomical parts. A design tool has been developed inorder to assist the user in defining the muscles in a natural manner by sketching their shapedirectly on the already existing bones and other muscles. The dynamics of the face motion iscomputed through a position-based schema ensuring real-time performance, control and robustness.Experiments demonstrate that through this model it can be effectively synthesizedrealistic expressive facial animation on different input face models in real-time on consumerclass platforms.The second method for automatically achieving animation consists in a novel facial motioncloning technique. It is a purely geometric algorithm and it is able to transfer the motionfrom an animated source face to a different target face mesh, initially static, allowing to reusefacial motion from already animated virtual heads. Its robustness and flexibility are assessedover several input data sets.
New 3D Scanning Techniques for Complex Scenes
(Chen, Tongbo, 2008-12-02) Chen, Tongbo
This thesis presents new 3D scanning methods for complex scenes, such as surfaceswith fine-scale geometric details, translucent objects, low-albedo objects,glossy objects, scenes with interreflection, and discontinuous scenes.Starting from the observation that specular reflection is a reliable visual cuefor surface mesostructure perception, we propose a progressive acquisition systemthat captures a dense specularity field as the only information for mesostructurereconstruction. Our method can efficiently recover surfaces with fine-scale geometricdetails from complex real-world objects.Translucent objects pose a difficult problem for traditional optical-based 3Dscanning techniques. We analyze and compare two descattering methods, phaseshiftingand polarization, and further present several phase-shifting and polarizationbased methods for high quality 3D scanning of translucent objects.We introduce the concept of modulation based separation, where a high frequencysignal is multiplied on top of another signal. The modulated signal inheritsthe separation properties of the high frequency signal and allows us to remove artifactsdue to global illumination. Thismethod can be used for efficient 3D scanningof scenes with significant subsurface scattering and interreflections.
Shape Deformations Based on Vector Fields
(Freiherr von Funck, Wolfram Alexander, 2008-12-22) Freiherr von Funck, Wolfram Alexander
This thesis explores applications of vector field processing to shape deformations. We present a novel method to construct divergence-free vector fields which are used to deform shapes by vector field integration. The resulting deformation is volume-preserving and no self-intersections occur. We add more controllability to this approach by introducing implicit boundaries, a shape editing method which resembles the well-known boundary constraint modeling metaphor. While the vector fields are originally defined in space, we also present a surface-based version of this approach which allows for more exact boundary selection and deformation control. We show that vectorfield-based shape deformations can be used to animate elastic motions without complex physical simulations. We also introduce an alternative approach to exactly preserve the volume of skinned triangle meshes. This is accomplished by constructing a displacement field on the mesh surface which restores the original volume after deformation. Finally, we demonstrate that shape deformation by vector field integration can also be used to visualize smoke-like streak surfaces in dynamic flow fields.
The Visible Vortex Interactive Analysis and Extraction of Vortices in Large Time-dependent Flow
(Fuchs, Sept 2008) Fuchs, Raphael
Computersimulation physikalischer und chemischer Prozesse sind ein essentielles Hilfsmittelzum Verständnis von Problemen aus dem Bereich der Strömungslehre geworden. Mit aktuellenSimulationspaketen ist es möglich für realistische Szenarien zeitabhängige Lösungenzu berechnen. Die berechneten Lösungen werden in großen Gitternetzen gespeichert, sindhäufig zeitabhängig und enthalten die multivariaten Ergebnisse der numerischen Simulation.Mit zunehmender Komplexität der Simulationsergebnisse entsteht die Notwendigkeitgeeignete Analyse- und Darstellungswerkzeuge zu verwenden, um aus den erzeugten Daten Erkenntnisse zu gewinnen und die gegebene Fragestellung lösen zu können. Dafür sindneue Methoden und Algorithmen notwendig um wichtige Teile der Daten zu extrahieren,wichtige Strukturen in der Strömung zu erkennen und das Strömungsverhalten interaktivanalysieren zu können.Das Hauptmotiv dieser Arbeit ist die Erweiterung vonWirbelkriterien auf zeitabhängigeLösungen und die Kombination dieser Wirbelkriterien mit den Methoden der interaktivenvisuellen Analyse. Um ein Verständnis für die Simulationsergebnisse zu entwickeln ist esnotwendig Attribute der Simulation miteinander vergleichen und mit größeren Strukturenwie Strömungswirbeln in Verbindung bringen zu können. Es wird gezeigt, wie automatischeWirbeldetektoren und interaktive Analyse kombiniert werden um sowohl Erkennung alsauch Analyse von Wirbelstrukturen zu verbessern.Indem klassische Wirbeldetektoren erweitert und in die visuelle Analyse integriert werden,ist es möglich die Auswirkung von Wirbeln auf wichtige Attribute der Flüssigkeit unddie Entwicklung der Strömung zu verstehen. Wir zeigen anhand von praktischen Beispielenaus dem Bereich der Motorentwicklung, dass Wirbel einen entscheidenden Einfluss aufanwendungskritische Variablen der Strömung haben können und demonstrieren wie interaktivevisuelle Analyse helfen kann diesen Einfluss einzuschätzen. Weiterhin besprechenwir wie zeitabhängige Größen in die Wirbelbestimmung einfließen können um die Qualitätder gefundenen Merkmale zu verbessern. - Computational simulation of physical and chemical processes has become an essential toolto tackle questions from the field of fluid dynamics. Using current simulation packagesit is possible to compute unsteady flow simulations for realistic scenarios. The resultingsolutions are stored in large to very large grids in 2D or 3D, frequently time-dependent,with multi-variate results from the numeric simulation. With increasing complexity ofsimulation results, powerful analysis and visualization tools are needed to make sense ofthe computed information and answer the question at hand. To do this we need newapproaches and algorithms to locate regions of interest, find important structures in theflow and analyze the behavior of the flow interactively.The main motives of this thesis are the extension of vortex detection criteria to unsteadyflow and the combination of vortex detectors with interactive visual analysis. To developan understanding for the simulation results it is necessary to compare attributes of thesimulation to each other and to be able to relate them to larger structures such as vortices.It is shown how automatic feature detection algorithms can be combined with interactiveanalysis techniques such that both detection and analysis benefit.By extending and integrating vortex detectors into the process of visual analysis, itbecomes possible to understand the impact of vortex structures on the development ofthe flow. Using real-world examples from the field of engine design we discuss how vortexstructures can have critical impact on the performance of a prototype. We illustrate howinteractive visual analysis can support prototype design and evaluation. Furthermore, weshow that taking the unsteady nature of the flow into account improves the quality of theextracted structures.
Contours and contrast
(Smith, Kaleigh, 2008-12-11) Smith, Kaleigh
Contrast in photographic and computer-generated imagery communicates colour and lightness differences that would be perceived when viewing the represented scene. Due to depiction constraints, the amount of displayable contrast is limited, reducing the image's ability to accurately represent the scene. A local contrast enhancement technique called unsharp masking can overcome these constraints by adding high-frequency contours to an image that increase its apparent contrast. In three novel algorithms inspired by unsharp masking, specialized local contrast enhancements are shown to overcome constraints of a limited dynamic range, overcome an achromatic palette, and to improve the rendering of 3D shapes and scenes. The Beyond Tone Mapping approach restores original HDR contrast to its tone mapped LDR counterpart by adding highfrequency colour contours to the LDR image while preserving its luminance. Apparent Greyscale is a multi-scale two-step technique that first converts colour images and video to greyscale according to their chromatic lightness, then restores diminished colour contrast with high-frequency luminance contours. Finally, 3D Unsharp Masking performs scene coherent enhancement by introducing 3D high-frequency luminance contours to emphasize the details, shapes, tonal range and spatial organization of a 3D scene within the rendering pipeline. As a perceptual justification, it is argued that a local contrast enhancement made with unsharp masking is related to the Cornsweet illusion, and that this may explain its effect on apparent contrast.
Efficient Line and Patch Feature Characterization and Management for Real-time Camera Tracking
(Wuest, 2008) Wuest, Harald
One of the key problems of augmented reality is the tracking of the camera position andviewing direction in real-time. Current vision-based systems mostly rely on the detectionand tracking of fiducial markers. Some markerless approaches exist, which are based on3D line models or calibrated reference images. These methods require a high manualpreprocessing work step, which is not applicable for the efficient development and designof industrial AR applications.The problem of the preprocessing overload is addressed by the development of vision-basedtracking algorithms, which require a minimal workload of the preparation of referencedata.A novel method for the automatic view-dependent generation of line models in real-timeis presented. The tracking system only needs a polygonal model of a reference object,which is often available from the industrial construction process. Analysis-by-synthesistechniques are used with the support of graphics hardware to create a connection betweenvirtual model and real model.Point-based methods which rely on optical flow-based template tracking are developedfor the camera pose estimation in partially known scenarios. With the support of robustreconstruction algorithms a real-time tracking system for augmented reality applicationsis developed, which is able to run with only very limited previous knowledge about thescene. The robustness and real-time capability is improved with a statistical approach fora feature management system which is based on machine learning techniques.III
Advanced Methods for Relightable Scene Representations in Image Space
(Fuchs, Martin, 2008-12-15) Fuchs, Martin
The realistic reproduction of visual appearance of real-world objectsrequires accurate computer graphics models that describe the opticalinteraction of a scene with its surroundings. Data-driven approachesthat model the scene globally as a reflectance field function in eightparameters deliver high quality and work for most material combinations,but are costly to acquire and store. Image-space relighting, whichconstrains the application to create photos with a virtual, fix camerain freely chosen illumination, requires only a 4D data structure toprovide full fidelity.This thesis contributes to image-space relighting on four accounts: (1)We investigate the acquisition of 4D reflectance fields in the contextof sampling and propose a practical setup for pre-filtering ofreflectance data during recording, and apply it in an adaptive samplingscheme. (2) We introduce a feature-driven image synthesis algorithm forthe interpolation of coarsely sampled reflectance data in software toachieve highly realistic images. (3) We propose an implicit reflectancedata representation, which uses a Bayesian approach to relight complexscenes from the example of much simpler reference objects. (4) Finally,we construct novel, passive devices out of optical components thatrender reflectance field data in real-time, shaping the incidentillumination into the desired image.
Evaluation and Enhancement of HDR Image Appearance on Displaysof Varying Dynamic Range
(Yoshida, Akiko, 2008-12-16) Yoshida, Akiko
The increasingly widespread availability of high dynamic range (HDR) technology has led to active study of the characteristics of the human visual system (HVS) in terms of brightness, lightness, contrast, and color perception and the application of the results of these studies to computer graphics. Because the development of HDR technology gives us display devices with much broader dynamic range for both high and low luminances, it is especially important to revise the models of HVS for the luminance ranges which are not covered by classical psychophysics, but required by the new HDR technology.In this dissertation, we focus on the evaluation and enhancement of the appearance of HDR images as reproduced on low dynamic range (LDR) media. First, we conducted a psychophysical experiment on seven tone mapping operators (TMOs)to assess how tone mapped images are perceived differently by human observers and to find out which attributes of image appearance account for these differences. The results show qualitative differences in TMOs, however, it also turned out that it was hard to choose the consistently best algorithm in terms of the fidelity of tone mapped images to real-world scenes. Based on this result, we conducted another series of psychophysical experiments with a generic easy to understand TMO. This experiment focused on three parameters - brightness, contrast, and color saturation - in order to measure user preference for, and fidelity of, tone mapped HDR images across several types of emulated limited dynamic range displays. The results of this study provide novel guidance for creating more advanced TMO designs.To evaluate the enhancement of image appearance, we conducted two series of experiments on contrast and brightness enhancement. A non-linear change in perceived contrast with respect to given physical contrast and a different adaptation luminance levels were measured by contrast scaling and contrast discrimination threshold experiments on HDR displays. These results lead to a model of just noticeable difference (JND) unit which provides uniformly changing perceived contrast in complex images. Finally, brightness enhancement caused by the glare illusion was investigated. We employed two profiles to evoke the glare illusion: a point spread function (PSF) of the human eye and a Gaussian kernel. The outcome of this study shows that the glare illusion increases the perceived luminance (brightness) by 20 - 35% for both convolution methods of a PSF and a Gaussian kernel. This means that faithful simulation of the human eye optics, which has been proposed before, is not necessary to achieve a strong brightness enhancement of the glare illusion because the Gaussian kernel, which has no theoretical justification in human perception, evokes the brightness enhancement at the same or higher strength than the PSF.
On Generalized Barycentric Coordinates and Their Applications in Geometric Modeling
(2008) Langer, Torsten
Generalized barycentric coordinate systems allow us to express the position of a point in space with respect to a given polygon or higher dimensional polytope. In such a system, a coordinate exists for each vertex of the polytope such that its vertices are represented by unit vectors ei (where the coordinate associated with the respective vertex is 1, and all other coordinates are 0). Coordinates thus have a geometric meaning, which allows for the simpli cation of a number of tasks in geometry processing. Coordinate systems with respect to triangles have been around since the 19th century, and have since been generalized; however, all of them have certain drawbacks, and are often restricted to special types of polytopes. We eliminate most of these restrictions and introduce a de nition for 3D mean value coordinates that is valid for arbitrary polyhedra in ?3, with a straightforward generalization to higher dimensions. Furthermore, we extend the notion of barycentric coordinates in such a way as to allow Hermite interpolation and investigate the capabilities of generalized barycentric coordinates for constructing generalized Bézier surfaces. Finally, we show that barycentric coordinates can be used to obtain a novel formula for curvature computation on surfaces.
Path-line Oriented Visualization of Dynamical Flow Fields
(2008) Shi, Kuangyu
An effective visual representation of dynamical flow behavior is still a challenging problem of modern flow visualization. Path-lines are important characteristic curves of dynamical flow fields. In this thesis, we focus on the visual analysis of path-line behaviors and uncover the dynamical nature of a flow field. We propose a topological segmentation of periodic 2D time-dependent vector fields based on asymptotic path-line behaviors. A flow domain is classified into different areas based on the converging or diverging path-line behaviors relating to the identified critical path-lines. We also offer an alternative algorithm to extract the separation surfaces of the path-line oriented topological structure. For the interactive visual analysis of fluid motion, we propose an information visualization based approach to explore the dynamical flow behaviors. Attributes associated with path-lines are identified and analyzed and the interesting features or structures are extracted and visualized with human interaction. We also investigate the property transport phenomenon and propose an approach to visualize the finite-time transport structures of property advection which is similar to carry out a line integral convolution over physical properties along path-lines. We demonstrate our approaches on a number of applications and present some interesting results.

Browse

Recent Submissions

Results Per Page

Sort Options