Styled Handwritten Text Generation (HTG) has received significant attention in recent years, propelled by the success of learning-based solutions employing GANs, Transformers, and, preliminarily, Diffusion Models. Despite this surge in interest, there remains a critical yet understudied aspect - the impact of the input, both visual and textual, on the HTG model training and its subsequent influence on performance. This work extends the VATr (Pippi et al. 2023) Styled-HTG approach by addressing the pre-processing and training issues that it faces, which are common to many HTG models. In particular, we propose generally applicable strategies for input preparation and training regularization that allow the model to achieve better performance and generalization capabilities. Moreover, in this work, we go beyond performance optimization and address a significant hurdle in HTG research - the lack of a standardized evaluation protocol. In particular, we propose a standardization of the evaluation protocol for HTG and conduct a comprehensive benchmarking of existing approaches. By doing so, we aim to establish a foundation for fair and meaningful comparisons between HTG strategies, fostering progress in the field.

BACKGROUND: Over the last years, substantial effort has been put into enhancing our arsenal in fighting epidemics from both technological and theoretical perspectives with scientists from different fields teaming up for rapid assessment of potentially urgent situations. This paper focusses on the computational aspects of infectious disease models and applies commonly available graphics processing units (GPUs) for the simulation of these models. However, fully utilizing the resources of both CPUs and GPUs requires a carefully balanced heterogeneous approach. RESULTS: The contribution of this paper is twofold. First, an efficient GPU implementation for evaluating a small-scale ODE model; here, the basic S(usceptible)-I(nfected)-R(ecovered) model, is discussed. Second, an asynchronous particle swarm optimization (PSO) implementation is proposed where batches of particles are sent asynchronously from the host (CPU) to the GPU for evaluation. The ultimate goal is to infer model parameters that enable the model to correctly describe observed data. The particles of the PSO algorithm are candidate parameters of the model; finding the right one is a matter of optimizing the likelihood function which quantifies how well the model describes the observed data. By employing a heterogeneous approach, in which both CPU and GPU are kept busy with useful work, speedups of 10 to 12 times can be achieved on a moderate machine with a high-end consumer GPU as compared to a high-end system with 32 CPU cores. CONCLUSIONS: Utilizing GPUs for parameter inference can bring considerable increases in performance using average host systems with high-end consumer GPUs. Future studies should evaluate the benefit of using newer CPU and GPU architectures as well as applying this method to more complex epidemiological scenarios.

Abstract In this paper we deal with paint‐on‐glass animation, which is a technique for making animated films by pushing slow‐drying paints from frame to frame directly under the camera. As artwork is continuously destroyed upon creating new frames, the animator is not able to rehearse or refine the animation afterwards. Furthermore, due to impracticable issues like how to stack up layers containing wet paint or how to overlay masks on the wet medium, one has to take both creative as technical decisions for each shot. Our approach consists of an interactive paint setup that physically simulates paint media. Together with a set of digital tools the artist is relieved from the difficult task of sustaining a constant frame‐to‐frame coherence while animating and is given the possibility to modify or undo earlier paint modifications. Regarding the interaction part, the artist stays in full control by employing a tangible interface. This allows artists to push pigment with their fingers, use special‐purpose digital brushes, or even to employ real life tools like paper tissue. Feedback from artists confirms the real‐life behaviour of stretching and smudging the paint as well as interacting with the setup, resulting in a natural and reality‐based methodology. Copyright © 2011 John Wiley & Sons, Ltd.

Advanced digital paint systems, based on accurate simulation models for the paint, brush and canvas interaction enable a virtual painting environment familiar to artists, but also allow for new ways of expressivity. The new FluidPaint co-located real-brush-input & canvas-display systems put extremely high demands on the real-time image processing and the reaction time between brush strokes and the rendering of the painting result on the screen. A new Smart Camera System-on-Chip (SoC) architecture is presented enabling real-time input of brush-strokes with real brushes in the FluidPaint digital painting system.

Abstract We propose a technique for fusing a bracketed exposure sequence into a high quality image, without converting to High dynamic range (HDR) first. Skipping the physically based HDR assembly step simplifies the acquisition pipeline. This avoids camera response curve calibration and is computationally efficient. It also allows for including flash images in the sequence. Our technique blends multiple exposures, guided by simple quality measures like saturation and contrast. This is done in a multiresolution fashion to account for the brightness variation in the sequence. The resulting image quality is comparable to existing tone mapping operators.

Abstract In this paper we describe the way in which emotion‐related data can efficiently be exchanged between participants in a large‐scale networked virtual environment. This type of metadata is extracted from real‐time captured video streams using off‐the‐shelf webcams and applied onto a two‐dimentional (2D) stylised avatar; thereby improving the immersion the user experiences while navigating and communicating in the virtual world. As emotion‐related data—once processed through the system—can be considered a specific type of state information, a generic networked virtual environment architecture can be used to distribute the information between participants. We have opted to extend the in‐house developed architecture for large‐scale virtual interactive communities (ALVIC‐NG) architecture to be able to process the information flows. We will show that the inclusion of this new type of information does not have a detrimental effect on the scalability of the system. Copyright © 2009 John Wiley & Sons, Ltd.

Abstract Many video sequences consist of a locally dynamic background containing moving foreground subjects. In this paper we propose a novel way of re‐displaying these sequences, by giving the user control over a virtual camera frame. Based on video mosaicing, we first compute a static high quality background panorama. After segmenting and removing the foreground subjects from the original video, the remaining elements are merged into a dynamic background panorama, which seamlessly extends the original video footage. We then re‐display this augmented video by warping and cropping the panorama. The virtual camera can have an enlarged field‐of‐view and a controlled camera motion. Our technique is able to process videos with complex camera motions, reconstructing high quality panoramas without parallax artefacts, visible seams or blurring, while retaining repetitive dynamic elements.

Abstract In this paper we argue for the concept of brushes and pigments enhanced with behavioural intelligence as a complement to established 2D computer animation techniques. Our approach allows the user to enrich and animate interactively created images by semi‐autonomously embedding procedural animation into them. This way, artistic skills of trained animators become within reach of everyone in the digital world, even to completely novice users. To exemplify this concept, we elaborate on several special‐purpose brushes and animated pigment for painting with animation. These include a programmable semi‐autonomous brush used to paint variations of strokes in successive frames (e.g. flames of fire), the moving‐dab brush which enables dabs of paint to follow a user‐defined path (e.g. smoke puffs) and animated pigment to produce secondary motion such as vibrant or waving paint textures. We believe our system offers a new fresh perspective on 2D computer aided animation production and associated tools. Copyright © 2008 John Wiley & Sons, Ltd.

We propose a technique for fusing a bracketed exposure sequence into a high quality image, without converting to HDR first. Skipping the physically-based HDR assembly step simplifies the acquisition pipeline. This avoids camera response curve calibration and is computationally efficient. It also allows for including flash images in the sequence. Our technique blends multiple exposures, guided by simple quality measures like saturation and contrast. This is done in a multiresolution fashion to account for the brightness variation in the sequence. The resulting image quality is comparable to existing tone mapping operators.

Abstract A novel approach is presented to efficiently render local subsurface scattering effects. We introduce an importance sampling scheme for a practical subsurface scattering model. It leads to a simple and efficient rendering algorithm, which operates in image space, and which is even amenable for implementation on graphics hardware. We demonstrate the applicability of our technique to the problem of skin rendering, for which the subsurface transport of light typically remains local. Our implementation shows that plausible images can be rendered interactively using hardware acceleration.

Abstract Existing work on applications for thin watery paint is mostly focused on automatic generation of painterly‐style images from input images, ignoring the fact that painting is a process that intuitively should be interactive. Efforts to create real‐time interactive systems are limited to a single paint medium and results often suffer from a trade‐off between real‐timeness and simulation complexity. We report on the design of a new system that allows the real‐time, interactive creation of images with thin watery paint. We mainly target the simulation of watercolor, but the system is also capable of simulating gouache and Oriental black ink. The motion of paint is governed by both physically based and heuristic rules in a layered canvas design. A final image is rendered by optically composing the layers using the Kubelka–Munk diffuse reflectance model. All algorithms that participate in the dynamics phase and the rendering phase of the simulation are implemented on graphics hardware. Images made with the system contain the typical effects that can be recognized in images produced with real thin paint, like the dark‐edge effect, watercolor glazing, wet‐on‐wet painting and the use of different pigment types. Copyright © 2005 John Wiley & Sons, Ltd.

Current watercolor painting systems produce convincing results, but trade this quality for a simulation that is much too slow for interactive use. We describe the implementation of a paper model that allows for real-time creation of watercolor images by simulating the mechanics of pigment and water throughout a three-layer paper model. In order to keep the simulation real-time while painting on large surfaces, we adopt a distributed paper canvas by dividing it in a grid of subpapers, each of which is delegated to a remote process

We present a novel approach to design artistic and believable trees in a cartoon-like style, which can be rendered by an animated camera to produce a convincing 3D-like experience. While computer assisted traditional animation is able to generate the desired effects, this approach still depends too much on the creation of many hand drawn images. Existing approaches fully depending on 3D geometries, on the other hand, give little artistic freedom and have difficulties to avoid artifacts popping up in successive frames. In order to provide good solutions to these difficulties, we present a hybrid (2.5D) framework, combining the advantages of both 2D and 3D approaches. From an underlying 3D geometry we get the necessary information to obtain an acceptable level of 3D behavior and a good frame-to-frame coherence. In the same framework, 2D artistic input is employed to obtain any desired 'look', both of the rendering and of the animation.

A novel approach is presented to efficiently render local subsurface scattering effects. We introduce an important sampling scheme for a practical subsurface scattering model. It leads to a simple and efficient rendering algorithm, which operates in image-space, and which is even amenable for implementation on graphics hardware. We demonstrate the applicability of our technique to the problem of skin rendering, for which the subsurface transport of light typically remains local. Our implementation shows that plausible images can be rendered interactively using hardware acceleration.

No abstract available.

The usage of computing systems has evolved dramatically over the last years. Starting from a low level procedural usage, in which a process for executing one or several tasks is carried out, computers now tend to be used in a problem oriented way. Future computer usage will be more centered around particular services, and will not be focused on platforms or applications. These services should be independent of the technology used to interact with them. In this paper an approach will be presented which provides a uniform interface to such services, without any dependence on modality, platform or programming language. Through the usage of general user interface descriptions, presented in XML, and converted using XSLT, a uniform framework is presented for runtime migration of user interfaces. As a consequence, future services will become easily extensible for all kinds of devices and modalities. Special attention goes out to a component-based software development approach. Services represented by and grouped in components can offer a special interface for modal- and device-independent rendering. Components become responsible for describing their own possibilities and constraints for interacting. An implementation serving as a proof of concept, a runtime conversion of a joystick in a 3D virtual environment into a 2D dialog-based user interface, is developed.

Although computer assistance for traditional animation is gaining a lot of attention during recent years, it still has to cope with many limitations. Part of the current research focuses on employing full 3D input models, which are rendered and even animated in many different non-photorealistic (NPR) styles. Disadvantages are the need to create complicated 3D models and the many difficulties to achieve lively movements. Purely 2D approaches, on the other hand, need many elaborated single drawings. Getting perspective right and retaining volumes are a major problem in that approach to computer assisted traditional animation, due to the complete lack of 3D information. Unfortunately, the employed software is ignorant about the approximate 3D representation in the animator's mind. In this paper, we present a novel tool for traditional animation, based on an approximate 3D model. This tool helps retaining volumes and proportions, and ensures frame-to-frame coherence.

In their paper about how the duality between subdivision surface schemes leads to higher-degree continuity, Zorin and Schroder (2001) consider only quadrilateral subdivision schemes. The dual of a quadrilateral scheme is again a quadrilateral scheme, while the dual of a triangular scheme is a hexagonal scheme. In this paper we propose such a hexagonal scheme, which can be considered a dual to Kobbelt's (2000) Sqrt(3) scheme for triangular meshes. We introduce recursive subdivision rules for meshes with arbitrary topology, optimizing the surface continuity given a minimal support area. These rules have a simplicity comparable to the Doo-Sabin scheme: only new vertices of one type are introduced and every subdivision step removes the vertices of the previous steps. As hexagonal meshes are not encountered-frequently in practice, we describe two different techniques to convert triangular meshes into hexagonal ones.

This paper introduces a new method for automatic in-betweening in computer assisted traditional animation. The solution is based on novel 2.5D modelling and animation techniques within the context of a multi-level approach, starting with basic 2D drawing primitives (curves) at level 0, over explicit 2.5D modelling structures at level 1 and inclusion of 3D information by means of skeletons at level 2, to high-level deformation tools (and possibly other tools for supporting specific purposes such as facial expression) at level 3. The underlying methodologies are explained and implementation results are elucidated.

Recursive subdivision surfaces allow a lot of freedom in designing surfaces of arbitrary topology. However, tools to manipulate them are still not as powerful as existing tools for established modeling paradigms such as B-spline surfaces. Furthermore, the most well behaved and most widely used schemes are approximating schemes, not interpolating their initial control points. We describe a new modeling paradigm providing local interpolation together with interactive normal and tension control to the approximating loop scheme, without interfering with its uniform subdivision rules. These tools are investigated for vertices both in the interior and at the border.

Voice over IP (VoIP) can be used in a wide variety of applications, all having different requirements. We present JVOIPLIB and JRTPLIB, a VoIP library and an RTP library respectively. Together they make it possible to easily add VoIP to various types of applications. Both libraries are written in an object-oriented style in C++, are open-source and are both very extensible. Several measures have been taken to allow good synchronization between the communicating parties.

Recursive subdivision schemes have become one of the most important paradigms to model 3D surfaces of arbitrary topology used in computer graphics applications. A number of researchers, both with a mathematical background and from the computer graphics community, have added-and still are adding-different algorithms and features to further improve their capabilities. This paper describes a new modeling tool, providing the possibility to locally choose an interpolating variant of the conventionally approximating Catmull-Clark (1978) subdivision scheme. Our approach combines the advantages of approximating schemes with the precise control of interpolating schemes. Unlike other solutions that mostly narrow down to locally change the weighting factors of the subdivision scheme, we keep the underlying uniform scheme intact. Our method is based upon introducing additional control points on well-chosen locations, with optional interactive user control over the tangent plane (or surface normal) and the tension of the surface near the interpolating control points. Although this paper is concentrating on the Catmull-Clark scheme, the proposed techniques can be extended to other subdivision schemes.

Conventional educational material is ever more complemented with computer based multimedia material. In order to make this material available to teachers and students in a structured manner, we developed a multimedia database and accompanying tools for creating, manipulating and formatting the teaching content (K. Coninx et al., 1997). Recently, we expanded this educational multimedia database with the functionality to support streamed video as well. Given the vast amounts of data that needs to be stored and transmitted when dealing with streamed video retrieval, we opted to realise a distributed video server. Physically, the distributed server consists of a set of Windows NT PCs, which are interconnected utilising a broadband ATM network. The paper reports on the main implementation aspects of the system.

Virtual environments and other applications of computer-generated scenes are emerging. This increases the demand for tools for efficient creation of the virtual worlds and interaction with the components in the virtual world. Introduction of immersiveness in the modeling process seems promising, but a number of issues related to immersive modeling still have to be investigated. The authors present a framework for an immersive modeling system with a hybrid 2D/3D user interface. The designer is immersed in the design space, which improves his awareness of the design objects under consideration. 2D and 3D human-computer interaction techniques are combined in the user interface to improve the efficiency of the modeler. While 3D interaction is natural for a number of design tasks, 2D interaction is particularly helpful for interaction with menus and dialog boxes, for precise manipulations and for editing operations that are subject to constraints.

The European project ELECTRA is concerned with the introduction of multimedia / telematics support to enhance the accessibility (local, regional and international) of learning material and to promote collaborative work. For the purpose of this paper we concentrate on two workpackages in the Limburg University Centre. The Interactive Multimedia Server workpackage focuses on the design of a generic platform for the integration of different types of information related to the learning process. Special attention is given to management, integration and accessibility of study material in an electronic environment. The other workpackage investigates the use of the interactive multimedia server in Medicine education.

This paper is about a 2 1/2 -D animation system, termed RUFIAS (from RUbbery FIgure Animation System), that is utilized for the creation of computer animation sequences which have the look and feel of traditional hand-drawn animation productions. RUFIAS provides a user interface where various basic methods of animation specification are provided, of which an approximated dynamic motion specification technique and motion curve based animation are the most important, and where various character drawing methods are provided, of which parameterized outline drawing and free-form skeleton object drawing are the most important. RUFIAS is implemented on state-of-the-art graphics workstations, allowing an interactive (as well as real-time) creation of animation sequences by means of direct manipulation.

This paper presents new techniques for speeding up the ray tracing algorithm. By projecting the scene to be ray traced onto perpendicular planes, subdividing the planes with a 2D subdivision scheme (regular grid- and quadtree structures) and traversing these projected scenes throughout the subdividing structures, the methods reported upon exploit the advantages of the 3D voxel- and octree- based acceleration techniques, without having their disadvantage of large memory consumption. Utilizing bounding rectangular extents around the primitives constituting the scene, the techniques covered can be beneficial for many different types of primitives to be ray traced. Results given for the regular grid- and quadtree-based techniques illustrate that even on single-processor machines it becomes viable to ray trace complex scenes --we tested the algorithms presented with scenes containing up to one million primitives-- in an affordable time (i.e. in the order of minutes rather than hours or days).

Abstract In the visual programming community, many interesting graphical metaphors have been reported upon for representing computer programs graphically. Most of them have a 2D or 2.5D appearance on the screen in order to reflect the inherent multi‐dimensionality of the programming constructs being represented. By going into a three‐dimensional representation, this reflection can go a stepfurther. With ever increasing3D graphics rendering capabilities on todays computers, it moreover becomes feasible to extend the dimensionality of the program (and data structure) depiction. We follow this approach by realizing 3D graphical programming techniques within CAEL, our interactive Computer Animation Environment Language. The paper elucidates how several concepts, traditionally found within the Virtual Environments area, can be utilized in the realization of three‐dimensional Programming Environments.

This paper is about ExTwAnPaSy, an Extensible Two-dimensional Animation/Paint System. ExTwAnPaSy is designed as a testbed for integrating various animation and painting techniques. Utilizing a uniform internal representation (and accompanying operations on it) for all geometric primitives, ExTwAnPaSy can integrate direct manipulation-, key-framing based-, as well as procedurally defined animation techniques. A key item of the system is that it can be expanded in a modular way using the standard programming environment feature of dynamic linking. By making the uniform internal representation accessible externally (by means of an abstract data type), a close interaction between the basic system and the added functionalities is guaranteed.>