Italian Chapter Conference 2025 - Smart Tools and Apps in Graphics

Permanent URI for this collection

https://diglib.eg.org/handle/10.2312/3607250

Browse

Now showing 1 - 20 of 26

Generalizing Shape-from-Template to Topological Changes
(The Eurographics Association, 2025) Manogue, Kevin; Schang, Tomasz M.; Kuş, Dilara; Müller, Jonas; Zachow, Stefan; Sengupta, Agniva; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
Reconstructing the surfaces of deformable objects from correspondences between a 3D template and a 2D image is well studied under Shape-from-Template (SfT) methods; however, existing approaches break down when topological changes accompany the deformation. We propose a principled extension of SfT that enables reconstruction in the presence of such changes. Our approach is initialized with a classical SfT solution and iteratively adapts the template by partitioning its spatial domain so as to minimize an energy functional that jointly encodes physical plausibility and reprojection consistency. We demonstrate that the method robustly captures a wide range of practically relevant topological events including tears and cuts on bounded 2D surfaces, thereby establishing the first general framework for topological-change-aware SfT. Experiments on both synthetic and real data confirm that our approach consistently outperforms baseline methods.
LiD2LOD: Generating LOD1 Urban Models from Airborne LiDAR
(The Eurographics Association, 2025) Sorgente, Tommaso; Moscoso Thompson, Elia; Romanengo, Chiara; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
The increasing availability of large-scale airborne LiDAR (Light Detection And Ranging) data related to urban scenarios requires the development of methods for transforming raw point clouds into structured 3D urban representations. The ability to generate accurate representations starting from raw data meets fundamental requirements in the fields of urban visualization, interactive simulation, and digital twins, providing a solid foundation for graphics and immersive reality applications. In this work, we present LiD2LOD, a framework for the automatic generation of Level of Detail 1 (LOD1) city models from LiDAR point clouds. Our tool can create both semantic models according to the CityGML standard, suitable for geospatial data integration, and lightweight triangular meshes, optimized for visualization and rendering. We test our approach on point clouds that represent historical cities characterized by complex morphology, thereby proving its scalability and robustness.
Algorithms in Geometric Deep learning and 3D AI: Theoretical Survey
(The Eurographics Association, 2025) Katturu, Vaibhav; Thind, Parampuneet Kaur; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
The study of shapes and geometric representations has long been central to Artificial Intelligence (AI). Early neural networks were limited to Euclidean domains such as images and sequences. The first extensions to non-Euclidean structures appeared in the 1990s and 2000s with recursive neural networks for hierarchical data and random walk-based graph methods. A major step forward came with spectral graph convolutional networks, which introduced convolution in the Fourier domain but faced scalability issues. Spatial methods later enabled more practical graph neural networks (GNNs). In parallel, 3D vision advanced with point cloud models such as PointNet and DGCNN, and mesh-based approaches like Geodesic CNN and MeshCNN, driving progress in classification, segmentation, and reconstruction. As algorithms in geometric deep learning and 3D AI expand, the field has grown both powerful and complex. This paper categorizes major algorithmic families, surveys key datasets across Euclidean and non-Euclidean domains, and highlights emerging advances and open research challenges.
Learning to Wait: Preventing Global Congestion from Local Observations in Real-Time Crowd Navigation
(The Eurographics Association, 2025) Ruprecht, Irena; Michelic, Florian; Preiner, Reinhold; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
We present a real-time crowd simulation approach based on reinforcement learning (RL), addressing congestion prevention in confined spaces. We learn a local navigation policy that uses compact, fast-to-compute per-agent observations of a small set of neighbors, including their desired directions. Alongside goal progress and inter-agent spacing, we reward agents for waiting when neighbors ahead pursue similar goals. This formulation fosters global self-organization from purely local interactions. Preliminary results show reduced congestion and consistent goal attainment for large crowds with hundreds of agents.
Smart Tools and Applications in Graphics - Eurographics Italian Chapter Conference: Frontmatter
(The Eurographics Association, 2025) Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
Virtual Crash Testing of Palletized Loads through Physics-Based Simulation
(The Eurographics Association, 2025) Boroni Grazioli, Matteo; Musoni, Pietro; Iotti, Eleonora; Palú, Alessandro Dal; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
Crash tests on loaded pallets are essential to ensure cargo stability during transport, as sudden accelerations or decelerations can cause loads to shift, topple, or become damaged. However, physical testing is costly, time-consuming, and destructive. To address these challenges, we present a virtual approach using physics-based simulation to model pallet behavior under realistic transport scenarios. In our method, both boxes and the stretch wrap are simulated using Extended Position-Based Dynamics (XPBD). Boxes are treated as rigid bodies, while distance constraints applied to the outer layer of the stack reproduce the compressive forces of real wrapping. Wrap failure is modeled by removing overstretched constraints, enabling the simulation of tearing events. Moreover, the widespread use of plastic wrapping has driven research into eco-friendly alternatives. Our simulation provides a safe, cost-effective way to evaluate such materials, supporting the study of pallet crash scenarios and laying the foundation for future improvements in physical realism, contact handling, and validation against real-world test data.
Combining CNN Feature Extraction and Kolmogorov-Arnold Networks Regression for Procedural 3D Shape Generation
(The Eurographics Association, 2025) Manfredi, Gilda; Capece, Nicola; Erra, Ugo; Gruosso, Antonio; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
Generating 3D objects with complex, nonlinear shapes directly from images is still an open research area. To address this problem, several state-of-the-art methods use Deep Learning (DL) to predict a set of parameters from images, which are then used to generate the 3D geometry, leveraging the characteristics of procedural modeling. Recently, Kolmogorov-Arnold Networks (KANs) have emerged as an alternative to traditional Multilayer Perceptrons (MLPs) in DL, and have been successfully integrated into architectures such as Convolutional Neural Networks (CNNs), Graph Neural Networks, and Transformers. In this work, we propose a DL architecture consisting of a hybrid CNN-KAN network for parametric 3D model generation from images. The model combines the ability of KANs to capture complex nonlinear functions with the strong visual feature extraction capabilities of CNNs. The method is evaluated using both quantitative error metrics and qualitative visualizations comparing predicted shapes with ground truth, and its performance is compared against a more standard CNN-MLP architecture.
Integrating Multi-Modal Solutions for Personalized and Accessible VR Museum Experiences
(The Eurographics Association, 2025) Bonino, Brigida; Giannini, Franca; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
Virtual Reality offers the opportunity to provide immersive and engaging experiences that can overcome physical, cultural, and economic barriers. However, defining and implementing principles and methodologies for accessible and inclusive virtual experiences remains a challenge, particularly because such systems must adapt to users with diverse needs related to age, culture, temporary physical conditions, or motor and sensory impairments. This work presents a virtual museum application that integrates multiple functionalities and provides options to address different physical and sensory issues, aiming to adapt the experience as much as possible to each single user. The system is designed to operate in a highly automatic way, minimizing user effort and stress while ensuring a comfortable and engaging experience.
Game mechanics and interaction for mixed reality and gamification
(The Eurographics Association, 2025) Beni, Andrea De; Castellani, Umberto; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
Realistic and effective interaction plays a very important role for extended reality scenarios, especially when a full integration between the real and synthetic world is required. Augmented reality (AR) frameworks still provide unsatisfactory performance due to the unreliability of the real-time AR systems and the lack of a systematic study on the user experience in the mixed reality environment. Gaming is a strong potential technique to make AR effective to combine immersion, engagement, and (social) interaction. However, the most of current AR games involve standard interaction mechanics that have been designed for generic games without considering the peculiarity of the augmented reality paradigm. In this work we address how to fill this gap between game and mixed reality. Our contribution is twofold. First, we introduce a practical and reliable mixed reality framework that adopt a model-based and marker-less approach to compute the camera matching procedure. Second, we propose and evaluate several game mechanics properly designed to work on the mixed reality scenarios. We employed our methods for cultural heritage application showing the effectiveness of the proposed mixed reality mechanics in improving the knowledge of archaeological findings in a playful way (i.e., gamification).
VOLMAP Viewer: a Web-Based Interactive Visual Tool to Explore Volume Maps
(The Eurographics Association, 2025) Giacalone, Alberto; Mulas, Dylan; Meloni, Federico; Cherchi, Gianmarco; Livesu, Marco; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
VOLMAP Viewer is a web-based tool for the visual assessment and analysis of maps between tetrahedral meshes. Users can load models directly in the browser and manage them with data structures optimized for interactive visualization. They can display and interact in real-time with both source and target domains, check map validity, and compute map quality according to a variety of popular distortion metrics. Fundamental interaction features such as volume slicing and synchronized picking are also possible: users can select any tetrahedron, highlighting the corresponding element on both domains and inspecting its associated map distortion. Thanks to its portability and interactive performances VOLMAP Viewer complements existing datasets and offline tools, supporting ongoing research on volume mapping algorithms.
ReCoGS: Real-time ReColoring for Gaussian Splatting scenes
(The Eurographics Association, 2025) Rutayisire, Lorenzo; Capodieci, Nicola; Pellacini, Fabio; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
Gaussian Splatting has emerged as a leading method for novel view synthesis, offering superior training efficiency and realtime inference compared to NeRF approaches, while still delivering high-quality reconstructions. Beyond view synthesis, this 3D representation has also been explored for editing tasks. Many existing methods leverage 2D diffusion models to generate multi-view datasets for training, but they often suffer from limitations such as view inconsistencies, lack of fine-grained control, and high computational demand. In this work, we focus specifically on the editing task of recoloring. We introduce a userfriendly pipeline that enables precise selection and recoloring of regions within a pre-trained Gaussian Splatting scene. To demonstrate the real-time performance of our method, we also present an interactive tool that allows users to experiment with the pipeline in practice. Code is available at https://github.com/loryruta/recogs.
TIGHT Intervals for Provably Correct Geometric Computation
(The Eurographics Association, 2025) Sichetti, Federico; Attene, Marco; Puppo, Enrico; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
Interval arithmetic is a practical method for robust computation, bridging the gap between fast, but inexact, floating-point arithmetic and slow, exact arithmetic, such as rational or arbitrary-precision. In this system, numbers are represented as intervals bounded by floating-point numbers, and operations are performed conservatively, guaranteeing that the resulting interval contains the exact mathematical result. We extend a fast C++ library for interval arithmetic by adding support for several transcendental functions. A key feature of our library is that all operations are correctly rounded, ensuring the resulting interval is the smallest floating-point interval that contains the true result. We demonstrate the library's effectiveness by applying it to complex non-polynomial problems, including surface-surface intersection and continuous collision detection for geometric primitives undergoing roto-translational motion.
Implicit Field-Based Stylization of 2D and 3D Liquid Animations
(The Eurographics Association, 2025) Stevenson-Regla, Rodrigo; Rohmer, Damien; Barthe, Loïc; Cani, Marie-Paule; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
Combining physically-based simulation with stylized visual effects not only requires to change the aspect of surfaces, but their shapes as well. We describe a new expressive rendering method for liquid animations, which can be used on top of any preexisting particle-based simulation. Our solution builds on visual particles that carry both water and air distributions, both evolving through particle history based on kinematics information from the simulation. These density fields are combined at each frame to create the implicit iso-surface of interest, rendered in adapted style. By defining series of visual particle states, we parametrize this model to capture the typical stylized geometry of water bodies used to highlight dynamic motion in paintings and cartoons, such as elongating droplets, concavities carved at the crest of breaking waves, and stylized air-water mixtures such as bubbles and foam. Regardless of the 2D or 3D nature of the input simulation, our solution maintains temporal coherence and ensures that water bodies keep an approximately constant surface in 2D, resp. or volume in 3D, over time. Finally, we conducted a user study to show the effectiveness of our method against state of the art AI-based tools, in a variety of animation scenarios where stylized shapes are needed.
Topographic Lifeline Maps
(The Eurographics Association, 2025) Smajic, Amel; Preiner, Reinhold; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
We introduce Topographic Lifeline Maps (TLMs), a visualization technique that combines a network of lifelines with topographic time maps to present genealogical graphs with both temporal detail and structural clarity. In TLMs, each individual's life is represented as a continuous curve that encodes birth, life events, and death in chronological order, while couples are depicted as parallel, merged lifelines. The layout is computed by a hybrid force model combining repulsion, temporally scaled springs, curvature-straightening, and octilinearization forces. Temporal interpretability is further enhanced through age-aware color encodings, with uncertain or missing dates distinguished visually to guide genealogical research. We demonstrate the expressiveness of TLMs on a large real-world dataset, illustrating diverse family stories, lifespans, and relational structures. A comparative user study against Topographic Attribute Maps (TAMs) shows that, while task performance is similar, participants consistently find TLMs more helpful across all analysis tasks. Our results highlight how mathematical layout formulations can advance genealogical visualization and suggest broader applicability to temporal-relational data beyond genealogy.
Structured Pattern Expansion with Diffusion Models
(The Eurographics Association, 2025) Riso, Marzia; Vecchio, Giuseppe; Pellacini, Fabio; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
Recent advances in diffusion models have significantly improved the synthesis of materials, textures, and 3D shapes. By conditioning these models on text or images, users can guide the generation, reducing the time required to create digital assets. In this paper, we address the synthesis of structured, stationary patterns, where diffusion models are generally less reliable and, more importantly, less controllable. Our approach leverages the generative capabilities of diffusion models specifically adapted to the pattern domain. It enables users to exercise direct control over the synthesis by expanding a partially hand-drawn pattern into a larger design while preserving the structure and details of the input. To enhance pattern quality, we fine-tune an image-pretrained diffusion model on structured patterns using Low-Rank Adaptation (LoRA), apply a noise rolling technique to ensure tileability, and utilize a patch-based approach to facilitate the generation of large-scale assets. We demonstrate the effectiveness of our method through a comprehensive set of experiments, showing that it outperforms existing models in generating diverse, consistent patterns that respond directly to user input. Code will be released at publication time at: https://github.com/marzia-riso/structured_pattern_expansion.
Learning to Predict Aboveground Biomass from RGB Images with 3D Synthetic Scenes
(The Eurographics Association, 2025) Zuffi, Silvia; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
Forests play a critical role in global ecosystems by supporting biodiversity and mitigating climate change via carbon sequestration. Accurate aboveground biomass (AGB) estimation is essential for assessing carbon storage and wildfire fuel loads, yet traditional methods rely on labor-intensive field measurements or remote sensing approaches with significant limitations in dense vegetation. In this work, we propose a novel learning-based method for estimating AGB from a single ground-based RGB image. We frame this as a dense prediction task, introducing AGB density maps, where each pixel represents tree biomass normalized by the plot area and each tree's image area. We leverage the recently introduced synthetic 3D SPREAD dataset, which provides realistic forest scenes with per-image tree attributes (height, trunk and canopy diameter) and instance segmentation masks. Using these assets, we compute AGB via allometric equations and train a model to predict AGB density maps, integrating them to recover the AGB estimate for the captured scene. Our approach achieves a median AGB estimation error of 1:22kg=m2 on held-out SPREAD data and 1:94kg=m2 on a real-image dataset. To our knowledge, this is the first method to estimate aboveground biomass directly from a single RGB image, opening up the possibility for a scalable, interpretable, and cost-effective solution for forest monitoring, while also enabling broader participation through citizen science initiatives.
BASS-MLIC: a Novel Synthetic Dataset for Single-View Inverse Rendering Tasks on Cultural Heritage Artifacts
(The Eurographics Association, 2025) Righetto, Leonardo; Ullah, Shakir; Giachetti, Andrea; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
BASS-MLIC is a synthetic dataset created with Blender to support multi-light image processing and inverse rendering tasks. It features orthographic views of culturally significant surfaces rendered with realistic materials and includes rich ground truth annotations, such as normals, depth, shadows, materials, and BRDF parameters. These annotations enable evaluation across diverse tasks like relighting, Photometric Stereo, shadow-aware estimations, and BRDF fitting. Preliminary experiments highlight its practical utility.
Organising and Enriching Urban 3D Models for Digital Twin Applications
(The Eurographics Association, 2025) Pittaluga, Simone; Cabiddu, Daniela; Mortara, Michela; Spagnuolo, Michela; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
The creation of an urban digital twin for the city of Matera, including 3D geometric models of the city landscape and heterogeneous information coming from different data sources, highlighted the need for a graphical user interface capable of integrating complex 3D urban scenes with geo-referenced data and knowledge (e.g., from real-time sensors, administrative information systems, algorithmic analysis and simulation, manual annotation) and making them easily accessible to a range of users with different levels of expertise. From these requirements, a new tool Matera3D has been developed as a specialized software platform for visualizing, analyzing and documenting 3D urban models. The application integrates point cloud and triangle mesh management with interactive semantic annotation, CityGML compliant dictionaries, and basic morphological analyzes, such as shadow computation and street slope measurements, as an initial set of processing tools that will be extended in the future. The software organizes all project data, including geometry, annotations, metadata, and scalar fields embedded in the geometry, within a coherent folder structure to ensure interoperability and efficient sharing. The software can show data from different sources and of different types in an integrated 3D view to facilitate research and practical applications in urban planning, environmental assessment, and infrastructure management beyond the popular 2D (GIS) approach.
Geometric aware local optimization for robust primitive fitting
(The Eurographics Association, 2025) Ferraris, Andrea; Leveni, Filippo; Baieri, Daniele; Maggioli, Filippo; Melzi, Simone; Magri, Luca; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
The decomposition of 3D point clouds into meaningful geometric primitives is a longstanding challenge in Computer Vision and Computer Graphics. While recent advances in data-driven methods and neural representations have achieved significant progress in 3D reconstruction and abstraction, traditional primitive-based representations remain invaluable for tasks requiring interpretability, compactness, and robustness. This work introduces a novel framework for primitive decomposition in 2D and 3D point clouds, designed to cope with noise, outliers, and overlapping structures. Building upon traditional RANSACbased approaches, the proposed method integrates geometric priors to enhance its effectiveness in identifying interpretable and meaningful geometric primitives within complex data. Central to our approach is a novel geometric-aware inlier refinement step, which incorporates geometric constraints such as surface completeness and normal consistency. This refinement step is formulated as an optimization problem solved through the GRAPH-CUT algorithm. This optimization process penalizes excessive surface extensions and promotes coherence in normal orientations, ensuring that the refined inlier sets closely match the geometric structures the point cloud represents. Experiments on synthetic and real-world datasets validate the robustness and accuracy of the proposed method, demonstrating its ability to outperform state-of-the-art techniques in terms of both result quality and computational efficiency.
NeuroShape: exploiting neural architectures for shape analysis of ultrastructural 3D neuroscience morphologies
(The Eurographics Association, 2025) Shaffique, Humaira; Shah, Uzair; Alzubaidi, Mahmood; Schneider, Jens; Magistretti, Pierre Julius; Cali, Corrado; Househ, Mowafa; Agus, Marco; Comino Trinidad, Marc; Mancinelli, Claudio; Maggioli, Filippo; Romanengo, Chiara; Cabiddu, Daniela; Giorgi, Daniela
Recent advances in volume electron microscopy (EM) enable nanometric-scale 3D reconstructions of neural tissue, providing unprecedented opportunities for studying cellular and subcellular morphology in neuroscience. The geometry of structures such as nuclei, neurites, and organelles can encode phenotypic information relevant to both functional specialization and pathological conditions, and thus represents a valuable complement to connectivity-based approaches in connectomics. While previous studies relied on handcrafted descriptors and classical machine learning for morphology analysis, recent progress in deep learning for 3D shape understanding offers new opportunities to learn robust, task-specific representations directly from geometric data. In this work we present NeuroShape, a first exploration of modern deep learning methods for shape analysis of ultrastructural 3D neuroscience morphologies. We introduce two annotated datasets derived from EM reconstructions: one of nuclei envelopes, and one of neurites and neural organelles. We benchmark two state-of-the-art neural architectures for 3D geometry (DiffusionNet [SACO22] and Laplacian2Mesh [DWL∗24]) and compare them against traditional feature-based descriptors previously used in neural morphology analysis. Our preliminary results highlight both the feasibility and the challenges of applying deep learning shape analysis techniques in this domain, and we release the datasets as a reference resource for future studies.

Browse

Browsing Italian Chapter Conference 2025 - Smart Tools and Apps in Graphics by Issue Date

Results Per Page

Sort Options