Eurographics Digital Library
Not a member yet
17121 research outputs found
Sort by
Generating 3D Hair Strips from Partial Strands using Diffusion Model
No full textAnimation-friendly hair representation is essential for real-time applications such as interactive character systems. While lightweight strip-based models are increasingly adopted as alternatives to strand-based hair for computational efficiency, creating such hair strips based on the hairstyle shown in a single image remains laborious. In this paper, we present a diffusion model-based framework for 3D hair strip generation using sparse strands extracted from a single portrait image. Our key idea is to formulate this task as an inpainting problem solved through a diffusion model operating in the UV parameter space of the head scalp. We parameterize both strands and strips on a shared UV scalp map, enabling the diffusion model to learn their correlations. We then perform spatial and channel-wise inpainting to reconstruct complete strip representations from partially observed strand maps. To train our diffusion model, we address the data scarcity problem of 3D hair strip models by constructing a large-scale strand-strip paired dataset through our adaptive clustering algorithm that converts groups of hair strands into strip models. Comprehensive qualitative and quantitative evaluations demonstrate that our framework effectively reconstructs high-quality hair strip models from an input image while preserving characteristic styles of strips. Furthermore, we show that the generated strips can be directly integrated into rigging-based animation workflows for real-time platforms such as games.Pacific Graphics Conference Papers, Posters, and Demos3D Reconstructio
NFGD: Neighborhood-Faithful Graph Drawing
No full textNeighborhood faithfulness metrics measure how faithfully the ground truth neighbors of vertices in a graph G are represented as the geometric neighbors of vertices in a drawing D of G. In this paper, we present NFGD, a post-processing algorithm for optimizing the neighborhood faithfulness of graph drawings. Experiments demonstrate the effectiveness of NFGD for computing neighbor-faithful drawings, on average 320% improvement over the popular graph drawing algorithms: 425% over Stress Majorization (SM) and 215% over force-directed algorithm Fruchterman-Reingold (FR). In particular, for scale-free graphs, NFGD-SM achieves 776% improvement over SM and NFGD-FR obtains 597% improvement over FR.EuroVis 2025 - Short PapersTechniques and Tool
''There was a scribe, a priest and a thief''. Testing the potential of language models for the creation of curatorial narratives in an archaeological museum
No full textIn the cultural heritage sector, Artificial Intelligence can aid in the creation of narratives by enhancing human creativity and assisting cultural heritage professionals in crafting and developing stories. In this work we focus specifically on how large language models can support rather than replace curatorial expertise. While AI can generate content, crafting compelling narratives requires human understanding of narrative structures, cultural context, and thematic coherence. We present a platform that helps curators create interactive stories for museums through AI assistance, developed as part of the CHANGES project at the Egyptian Museum in Turin. The platform enables curatorial narrative creation with selective LLM support, connects stories to museum collections via semantic annotation, and facilitates translation to venue-specific technical formats. Rather than having LLMs generate complete stories, curators construct the narrative framework while using LLMs to transform structured scene descriptions into polished prose. Our evaluation with three state-of-the-art models across 147 scenes and in a real use-case scenario shows that current LLMs can effectively complete this constrained creative task, though all outputs still require human refinement. This curator-driven approach ensures that generated narratives maintain the accuracy and scholarly standards essential for cultural heritage contexts while benefiting from AI's linguistic capabilities.Digital HeritageDigital Technologies for CHANGES (CHANGES SESSION) - Part
Single-Line Drawing Vectorization
No full textVectorizing line drawings is a repetitive, yet necessary task that professional creatives must perform to obtain an easily editable and scalable digital representation of a raster sketch. State-of-the-art automatic methods in this domain can create series of curves that closely fit the appearance of the drawing. However, they often neglect the line parameterization. Thus, their vector representation cannot be edited naturally by following the drawing order. We present a novel method for single-line drawing vectorization that addresses this issue. Single-line drawings consist of a single stroke, where the line can intersect itself multiple times, making the drawing order non-trivial to recover. Our method fits a single parametric curve, represented as a Bézier spline, to approximate the stroke in the input raster image. To this end, we produce a graph representation of the input and employ geometric priors and a specially trained neural network to correctly capture and classify curve intersections and their traversal configuration. Our method is easily extended to drawings containing multiple strokes while preserving their integrity and order.We compare our vectorized results with the work of several artists, showing that our stroke order is similar to the one artists employ naturally. Our vectorization method achieves state-of-the-art results in terms of similarity with the original drawing and quality of the vectorization on a benchmark of single-line drawings. Our method's results can be refined interactively, making it easy to integrate into professional workflows. Our code and results are available at https://github.com/tanguymagne/SLD-Vectorization.Computer Graphics ForumLines, Surfaces & Fields44
Evolution of Extreme Dust Events in 3D Environment
No full textDust is one of the main components of atmospheric particles in desert regions. The concentration, composition, and spatial distribution of these dust particles in the atmosphere vary over time and could significantly impact the weather, climatological conditions, radiative forcing and transfer, and ecosystem dynamics. Scientists and decision-makers are interested in analyzing the evolution of dust events (including their formation, dynamics, and interactions with the environment), understanding the main contributing factors and atmospheric conditions that intensify these events and lead to extreme dust events, examining the role of topographic features, and gaining insights into their relationship with global teleconnections. Scientists use highresolution simulation models (such as WRF-Chem) to simulate the prevalent atmospheric conditions centered around extreme events and examine the model outputs to understand such events better. These simulation datasets are extremely large in scale, depending on the spatial domain, spatiotemporal resolutions, simulation duration, and number of atmospheric parameters. They need a specialized environment that facilitates analyzing such datasets. To this end, we provide a 3D visualization system that facilitates the analysis of dust simulation model outputs and provides information about dust loading, transport, evolution, deposition, and intensification into an extreme event. This system also aids in understanding the interactions between different atmospheric parameters, the impact of terrain surface characteristics, and more, providing a holistic view of the dust events. We present a case study demonstrating the system's capabilities in analyzing extreme dust events and also include feedback from the domain experts, along with a discussion on future extensions.Workshop on Visualisation in Environmental Sciences (EnvirVis)Session
Enhancing South Slavic Cyrillic Manuscripts Research through a Digital Toolkit for Cyrillic Palaeography
No full textThis paper presents an innovative digital tool designed to support the analysis of medieval Cyrillic handwriting within a codicological and palaeographical framework. The Cyrillic Palaeography Toolkit (CyPaT), developed from a standardised descriptive model and integrated with the Repertorium of South Slavonic Manuscripts and Copyists (X-XIV cc.), offers a comprehensive environment for the description, processing, and comparative study of manuscript data. By enabling direct interaction with digitised images, CyPaT allows for detailed examination of script features such as letter proportions, stroke composition, and layout. As a free and open-source resource, it promotes collaborative research and the systematic identification of script features linked to specific periods, scriptoria, or individual scribes, particularly within the South Slavic Cyrillic tradition, marking a significant step forward in the development of Digital Slavic Palaeography as an emerging field of scholarly inquiry.Digital HeritagePoster
Modeling Audience NPC's Diversity for Enhanced Illusionary Sociality in Collective VR Concert Experiences
No full textVirtual reality (VR) live performances offer a novel style of experience through the deployment of three-dimensional cartoonlike performers and through spatiotemporally asynchronous participation. However, such experiences often lack the presence and atmospheric richness of physical live concerts. To address this gap, prior research has proposed audience non-player characters (NPCs) that emulate the collective characteristics of real-world audiences to enhance viewer engagement. In this study, we extend existing approaches by incorporating individual motion diversity to synthesize audience NPCs that form a heterogeneous and lifelike crowd. In addition to collective characteristics, such as music-driven motion changes and gradual synchronization, our modeling approach introduces agent-level diversity, even within the same motion categories. To assess the psychological impact of the proposed method, we developed a VR concert scenario featuring approximately 1,500 audience NPCs and conducted an empirical evaluation of the user experience. The findings indicate that enhanced motion heterogeneity can increase a viewer's sense of presence, engagement, perceived unity, and co-existence in a VR concert environment.ICAT-EGVE 2025 - International Conference on Artificial Reality and Telexistence and Eurographics Symposium on Virtual EnvironmentsSoun
The Multimedia Archive of the Academy of Fine Arts of Florence as a Tool for the Enhancement, Promotion, and Preservation of Artistic Heritage
No full textThe Multimedia Archive of the Academy of Fine Arts of Florence was born from the need to digitize and preserve the vast artistic heritage housed within the historic Florentine institution. The project, which began in 2022, aimed to identify a tool capable of enhancing and promoting cultural and artistic content through the integration of art and technological innovation. By examining contemporary society and the tools through which it accesses content and information such as smartphones and computers it has become evident that an easily accessible web platform constitutes one of the most effective means. Furthermore, new technologies enable the delivery of interactive multimedia content capable of meeting the needs of younger generations who are approaching the world of art for the first time.Digital HeritagePoster
From Interpretation to Immersion: XR and the Transformation of Fashion Heritage Storytelling
No full textBrand heritage is a fundamental element in shaping fashion brands' identity, encapsulating narratives of craftsmanship, culture, and legacy. With the rise of Extended Reality (XR) technologies, new opportunities emerge for preserving and communicating these narratives, redefining how brands engage with their audiences. This study explores how XR enhances fashion heritage storytelling by facilitating interactive and immersive experiences beyond traditional archival and marketing approaches. Grounded in Riviezzo definition of brand heritage activities and Mosca's heritage factors for brands' storytelling techniques, the research examines XR applications across five key dimensions: (1) historical evolution, (2) influential figures, (3) craftsmanship techniques, (4) reissues of iconic products, and (5) architectural spaces emblematic of brand identity. Through case studies, the paper demonstrates how XR transforms storytelling into storyliving, fostering deeper audience engagement with historical garments, artisanal techniques, and brand histories. Additionally, it envisions future scenarios where XR and AI-driven personalisation expand the possibilities of brand heritage, integrating multisensory interactivity and user-driven narratives. Positioning XR as an instrument that crosses between tradition and innovation, this study contributes to discussions on the phygital transformation of fashion, offering insights into experiential approaches to brand heritage preservation and engagement.Digital HeritagePhygital Worlds and XR in Cultural Heritag
Real-Time and Controllable Reactive Motion Synthesis via Intention Guidance
No full textWe propose a real-time method for reactive motion synthesis based on the known trajectory of an input character, predicting instant reactions using only historical, user-controlled motions. Our method handles the uncertainty of future movements by introducing an intention predictor, which forecasts key joint intentions to make pose prediction more deterministic from the historical interaction. The intention is later encoded into the latent space of its reactive motion, matched with a codebook that represents mappings between input and output. It samples from the categorical distribution for pose generation and strengthens model robustness through adversarial training. Unlike previous offline approaches, the system can recursively generate intentions and reactive motions using feedback from earlier steps, enabling real-time, long-term realistic interactive synthesis. Both quantitative and qualitative experiments show our approach outperforms other matching-based motion synthesis approaches, delivering superior stability and generalisability. In our method, the user can also actively influence the outcome by controlling the moving directions, creating a personalised interaction path that deviates from predefined trajectories.Computer Graphics ForumMajor Revision from Eurographics Conference44