Eurographics Digital Library
Not a member yet
17121 research outputs found
Sort by
Radiative Backpropagation with Non-Static Geometry
No full textRadiative backpropagation-based (RB) methods efficiently compute reverse-mode derivatives in physically-based differentiable rendering by simulating the propagation of differential radiance. A key assumption is that differential radiance is transported like normal radiance. We observe that this holds only when scene geometry is static and demonstrate that current implementations of radiative backpropagation produce biased gradients when scene parameters change geometry. In this work, we derive the differential transport equation without assuming static geometry. An immediate consequence is that the parameterization matters when the sampling process is not differentiated: only surface integrals allow a local formulation of the derivatives, i.e., one in which moving surfaces do not affect the entire path geometry. While considerable effort has been devoted to handling discontinuities resulting from moving geometry, we show that a biased interior derivative compromises even the simplest inverse rendering tasks, regardless of discontinuities. An implementation based on our derivation leads to systematic convergence to the reference solution in the same setting and provides unbiased RB interior derivatives for path-space differentiable rendering.Eurographics Symposium on RenderingDifferentiable Renderin
Reshaping Identities. How the Ugo andOlga Levi Foundation Navigates the Digital Age Making Sense of Its Heritage
No full textThe paper examines how the identity of a cultural institution is deeply connected with its territory and how both must adapt to contemporary socio-economic challenges. In this context, digital innovation emerges as a strategic lever for institutional transformation, reshaping identities, expanding outreach, and developing new models of engagement and accessibility. The case study of Ugo and Olga Levi Foundation, located in Venice, exemplifies how the ongoing digital transformation process is reshaping the institution's identity, generating both internal and external impacts. Historically, Palazzo Giustinian Lolin was a cultural hub thanks to its central location and the work of the Levi couple, who transformed their residence into a salon for music, art, and culture. After their death, it became the seat of the Ugo and Olga Levi Foundation, hosting events and research activities. Today, in a context of mass tourism and globalization, the Foundation seeks to maintain its territorial connection while pursuing new forms of cultural accessibility through the implementation of the LeviDigiLab, its digitization laboratory. The findings indicate that the digitization process has the potential to reshape the Foundation's identity, acting as a driver of its cultural and social regeneration. While this case reflects how one institution responds to its unique local conditions, further research is required to assess how such dynamics manifest in other urban contexts.Digital HeritageDigital Heritage, Tourism, and Sustainabilit
Tracing Intangible Heritage from Eurasia to Local Territories: Digital Projects of the I Deug-Su Centre
No full textThe Centro Studi Comparati I Deug-Su at the University of Siena promotes the application of Digital Humanities to Medieval Latin Philology. Its platforms-ALIM, CORIMU, ELA, and RAMMSES-offer digital editions of Latin texts, lexical resources, and multimedia archives. The Centre also coordinates programs in text encoding, digital editing, and archiving, enhancing interdisciplinary studies and access to Latin cultural heritage.Digital HeritagePoster
Versatile Physics-based Character Control with Hybrid Latent Representation
No full textWe present a versatile latent representation that enables physically simulated character to efficiently utilize motion priors. To build a powerful motion embedding that is shared across multiple tasks, the physics controller should employ rich latent space that is easily explored and capable of generating high-quality motion. We propose integrating continuous and discrete latent representations to build a versatile motion prior that can be adapted to a wide range of challenging control tasks. Specifically, we build a discrete latent model to capture distinctive posterior distribution without collapse, and simultaneously augment the sampled vector with the continuous residuals to generate high-quality, smooth motion without jittering. We further incorporate Residual Vector Quantization, which not only maximizes the capacity of the discrete motion prior, but also efficiently abstracts the action space during the task learning phase. We demonstrate that our agent can produce diverse yet smooth motions simply by traversing the learned motion prior through unconditional motion generation. Furthermore, our model robustly satisfies sparse goal conditions with highly expressive natural motions, including head-mounted device tracking and motion in-betweening at irregular intervals, which could not be achieved with existing latent representations.Computer Graphics ForumBringing Motion to Life: Motion Reconstruction and Control44
Developing a Controlled Experimental Space in VRChat: Comparing Embodiment in VRChat Users and Non-VRChat Users
No full textIn virtual reality, it is possible for users to feel as if the avatar were their own innate body, a phenomenon known as the sense of embodiment. In conventional laboratory experiments, the movements and observations that induce avatar embodiment have typically been limited to a short duration of about five minutes. In contrast, in social VR, users are considered to naturally engage in prolonged movements and observations, and it is possible that the sense of embodiment is strengthened through long-term play. In this study, we constructed an experimental space in VRChat that enables the controlled manipulation and measurement of the sense of embodiment, and we compared VRChat users with non-users. The results showed that there was no significant difference in the sense of embodiment depending on whether or not participants had prior VRChat experience. On the other hand, among VRChat users, it was revealed that the component of embodiment known as self-location became stronger as playtime increased. In addition, our findings confirmed that the control and measurement of embodiment can be conducted in VRChat in a manner comparable to laboratory experiments. By utilizing the experimental space developed in this study, it will be possible in the future to conduct various experiments that investigate VRChat users in the extended context of the environments in which they live.ICAT-EGVE 2025 - International Conference on Artificial Reality and Telexistence and Eurographics Symposium on Virtual EnvironmentsEmbodiment and Navigatio
GaussianMatch: Adaptive Learning Continuous Surfaces for Light Field Depth Estimation
No full textLight field (LF) depth estimation plays a vital role in computational imaging by reconstructing 3D structures from multiple viewpoints. However, images are merely discrete expressions of scenes due to the resolution constraints of cameras, leading to depth discontinuities and outliers-particularly in textureless or occluded regions, degrading reconstruction coherence. To address the challenges mentioned above, we propose GaussianMatch, a probabilistic depth estimation framework that models per-pixel depth as a learnable Gaussian distribution in continuous space. This scheme effectively alleviates the discretization problem of LF images by adaptively reconstructing continuous surfaces, while enabling uncertainty-aware optimization.Furthermore, the framework naturally fuses information among adjacent pixels and adapts each Gaussian's variance according to scene complexity, achieving robustness in both texture-rich and ambiguous regions. We further design GaussianNet, which regresses per-pixel Gaussian parameters and generates the final depth map via Gaussian accumulation. Extensive experiments on multiple LF benchmarks demonstrate that GaussianNet achieves state-of-the-art accuracy, with significant improvements in handling depth discontinuities and occlusions.Pacific Graphics Conference Papers, Posters, and DemosPoint Clouds & Gaussian Splattin
Weighted Feature Graph via Hierarchical Clustering
No full textIn computer graphics, mesh clustering is a key component of various applications such as shape matching or skinning weight computation, especially when using hierarchical clustering. Garland et al. [GWH01] proposed to build a hierarchy of clusters by simplifying the dual graph of the mesh. We extend their method to provide control over cluster shapes through a combination of error metrics. Additionally, we alleviate the challenging task of finding an optimal threshold (stopping criterion) by considering a weighted feature graph that incorporates persistent cluster information throughout the hierarchy.Eurographics 2025 - PostersPoster
FlairGPT: Repurposing LLMs for Interior Designs
No full textInterior design involves the careful selection and arrangement of objects to create an aesthetically pleasing, functional, and harmonized space that aligns with the client's design brief. This task is particularly challenging, as a successful design must not only incorporate all the necessary objects in a cohesive style, but also ensure they are arranged in a way that maximizes accessibility, while adhering to a variety of affordability and usage considerations. Data-driven solutions have been proposed, but these are typically room- or domain-specific and lack explainability in their design design considerations used in producing the final layout. In this paper, we investigate if large language models (LLMs) can be directly utilized for interior design. While we find that LLMs are not yet capable of generating complete layouts, they can be effectively leveraged in a structured manner, inspired by the workflow of interior designers. By systematically probing LLMs, we can reliably generate a list of objects along with relevant constraints that guide their placement. We translate this information into a design layout graph, which is then solved using an off-the-shelf constrained optimization setup to generate the final layouts. We benchmark our algorithm in various design configurations against existing LLM-based methods and human designs, and evaluate the results using a variety of quantitative and qualitative metrics along with user studies. In summary, we demonstrate that LLMs, when used in a structured manner, can effectively generate diverse high-quality layouts, making them a viable solution for creating large-scale virtual scenes. Code is available via the project webpage.Computer Graphics ForumShape It Til You Make It: Programs for 3D Synthesis44
Running Online User Studies with the reVISit Framework
No full textThere currently are two main approaches for running online user studies: experimenters can use commercial survey tools, which are easy to use but can be costly, hamper reproducibility, and have limitations for complex stimuli; or they can build custom software to run and instrument a study, which is a laborious and complex task. In this tutorial, we introduce participants to a new, open-source alternative: the reVISit study platform. Many studies quickly reach a level of complexity such that designers have not only to consider their stimuli and experimental tasks, but also the study UI, data hosting, participant recruiting, randomization, etc. ReVISit ameliorates these problems and allows study designers to focus more on the research questions and stimulus design. ReVISit removes the tedium of study design by providing built-in components that most studies will need. ReVISit uses a domain specific language to allow study designers to quickly create studies, and to deploy them as static websites that are publicly accessible. This tutorial will introduce reVISit to the visualization community and allow community members to get hands on experience with it through a series of practical examples. Throughout the tutorial, participants will improve on a study until they have developed and deployed a study of an interactive, fully instrumented data visualization.EuroVis 2025 - Panels and TutorialsTutorial
Certainly Uncertain: Reintroducing Uncertainty in Visualizations
No full textInformation Diffusion (ID) is shaped by uncertainty, yet most visualizations overlook it, leading to oversimplified or misleading interpretations. This work enhances two existing ID visualizations by integrating uncertainty through visual encodings within the original research goals. We are exploring how visualizing uncertainty might influence interpretation, including the potential for signal suppression or amplification. We discuss design alternatives and insights that apply to visualizing uncertainty in two existing visualization techniques. Future work directions are focusing on evaluating the designs and eliciting user feedback and comments on the interpretability and intuitiveness of the proposed uncertainty visualization encodings.EuroVis 2025 - PostersPoster