75 results
Search Results
Now showing 1 - 10 of 75
Item StyleBlend: Enhancing Style-Specific Content Creation in Text-to-Image Diffusion Models(The Eurographics Association and John Wiley & Sons Ltd., 2025) Chen, Zichong; Wang, Shijin; Zhou, Yang; Bousseau, Adrien; Day, AngelaSynthesizing visually impressive images that seamlessly align both text prompts and specific artistic styles remains a significant challenge in Text-to-Image (T2I) diffusion models. This paper introduces StyleBlend, a method designed to learn and apply style representations from a limited set of reference images, enabling content synthesis of both text-aligned and stylistically coherent. Our approach uniquely decomposes style into two components, composition and texture, each learned through different strategies. We then leverage two synthesis branches, each focusing on a corresponding style component, to facilitate effective style blending through shared features without affecting content generation. StyleBlend addresses the common issues of text misalignment and weak style representation that previous methods have struggled with. Extensive qualitative and quantitative comparisons demonstrate the superiority of our approach.Item Neural Geometry Processing via Spherical Neural Surfaces(The Eurographics Association and John Wiley & Sons Ltd., 2025) Williamson, Romy; Mitra, Niloy J.; Bousseau, Adrien; Day, AngelaNeural surfaces (e.g., neural map encoding, deep implicit, and neural radiance fields) have recently gained popularity because of their generic structure (e.g., multi-layer perceptron) and easy integration with modern learning-based setups. Traditionally, we have a rich toolbox of geometry processing algorithms designed for polygonal meshes to analyze and operate on surface geometry. Without an analogous toolbox, neural representations are typically discretized and converted into a mesh, before applying any geometry processing algorithm. This is unsatisfactory and, as we demonstrate, unnecessary. In this work, we propose a spherical neural surface representation for genus-0 surfaces and demonstrate how to compute core geometric operators directly on this representation. Namely, we estimate surface normals and first and second fundamental forms of the surface, as well as compute surface gradient, surface divergence and Laplace Beltrami operator on scalar/vector fields defined on the surface. Our representation is fully seamless, overcoming a key limitation of similar explicit representations such as Neural Surface Maps [MAKM21]. These operators, in turn, enable geometry processing directly on the neural representations without any unnecessary meshing. We demonstrate illustrative applications in (neural) spectral analysis, heat flow and mean curvature flow, and evaluate robustness to isometric shape variations. We propose theoretical formulations and validate their numerical estimates, against analytical estimates, mesh-based baselines, and neural alternatives, where available. By systematically linking neural surface representations with classical geometry processing algorithms, we believe this work can become a key ingredient in enabling neural geometry processing. Code is available via the project webpage.Item Lipschitz Pruning: Hierarchical Simplification of Primitive-Based SDFs(The Eurographics Association and John Wiley & Sons Ltd., 2025) Barbier, Wilhem; Sanchez, Mathieu; Paris, Axel; Michel, Élie; Lambert, Thibaud; Boubekeur, Tamy; Paulin, Mathias; Thonat, Theo; Bousseau, Adrien; Day, AngelaRendering tree-based analytical Signed Distance Fields (SDFs) through sphere tracing often requires to evaluate many primitives per tracing step, for many steps per pixel of the end image. This cost quickly becomes prohibitive as the number of primitives that constitute the SDF grows. In this paper, we alleviate this cost by computing local pruned trees that are equivalent to the full tree within their region of space while being much faster to evaluate. We introduce an efficient hierarchical tree pruning method based on the Lipschitz property of SDFs, which is compatible with hard and smooth CSG operators. We propose a GPU implementation that enables real-time sphere tracing of complex SDFs composed of thousands of primitives with dynamic animation. Our pruning technique provides significant speedups for SDF evaluation in general, which we demonstrate on sphere tracing tasks but could also lead to significant improvement for SDF discretization or polygonization.Item Deformed Tiling and Blending: Application to the Correction of Distortions Implied by Texture Mapping(The Eurographics Association and John Wiley & Sons Ltd., 2025) Wendling, Quentin; Ravaglia, Joris; Sauvage, Basile; Bousseau, Adrien; Day, AngelaThe prevailing model in virtual 3D scenes is a 3D surface, which a texture is mapped onto, through a parameterization from the texture plane. We focus on accounting for the parameterization during the texture creation process, to control the deformations and remove the cuts induced by the mapping. We rely on the tiling and blending, a real-time and parallel algorithm that generates an arbitrary large texture from a small input example. Our first contribution is to enhance the tiling and blending with a deformation field, which controls smooth spatial variations in the texture plane. Our second contribution is to derive, from a parameterized triangle mesh, a deformation field to compensate for texture distortions and to control for the texture orientation. Our third contribution is a technique to enforce texture continuity across the cuts, thanks to a proper tile selection. This opens the door to interactive sessions with artistic control, and real-time rendering with improved visual quality.Item D-NPC: Dynamic Neural Point Clouds for Non-Rigid View Synthesis from Monocular Video(The Eurographics Association and John Wiley & Sons Ltd., 2025) Kappel, Moritz; Hahlbohm, Florian; Scholz, Timon; Castillo, Susana; Theobalt, Christian; Eisemann, Martin; Golyanik, Vladislav; Magnor, Marcus; Bousseau, Adrien; Day, AngelaDynamic reconstruction and spatiotemporal novel-view synthesis of non-rigidly deforming scenes recently gained increased attention. While existing work achieves impressive quality and performance on multi-view or teleporting camera setups, most methods fail to efficiently and faithfully recover motion and appearance from casual monocular captures. This paper contributes to the field by introducing a new method for dynamic novel view synthesis from monocular video, such as casual smartphone captures. Our approach represents the scene as a dynamic neural point cloud, an implicit time-conditioned point distribution that encodes local geometry and appearance in separate hash-encoded neural feature grids for static and dynamic regions. By sampling a discrete point cloud from our model, we can efficiently render high-quality novel views using a fast differentiable rasterizer and neural rendering network. Similar to recent work, we leverage advances in neural scene analysis by incorporating data-driven priors like monocular depth estimation and object segmentation to resolve motion and depth ambiguities originating from the monocular captures. In addition to guiding the optimization process, we show that these priors can be exploited to explicitly initialize our scene representation to drastically improve optimization speed and final image quality. As evidenced by our experimental evaluation, our dynamic point cloud model not only enables fast optimization and real-time frame rates for interactive applications, but also achieves competitive image quality on monocular benchmark sequences. Our code and data are available online https://moritzkappel.github.io/projects/dnpc/.Item Versatile Physics-based Character Control with Hybrid Latent Representation(The Eurographics Association and John Wiley & Sons Ltd., 2025) Bae, Jinseok; Won, Jungdam; Lim, Donggeun; Hwang, Inwoo; Kim, Young Min; Bousseau, Adrien; Day, AngelaWe 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.Item Does 3D Gaussian Splatting Need Accurate Volumetric Rendering?(The Eurographics Association and John Wiley & Sons Ltd., 2025) Celarek, Adam; Kopanas, Georgios; Drettakis, George; Wimmer, Michael; Kerbl, Bernhard; Bousseau, Adrien; Day, AngelaSince its introduction, 3D Gaussian Splatting (3DGS) has become an important reference method for learning 3D representations of a captured scene, allowing real-time novel-view synthesis with high visual quality and fast training times. Neural Radiance Fields (NeRFs), which preceded 3DGS, are based on a principled ray-marching approach for volumetric rendering. In contrast, while sharing a similar image formation model with NeRF, 3DGS uses a hybrid rendering solution that builds on the strengths of volume rendering and primitive rasterization. A crucial benefit of 3DGS is its performance, achieved through a set of approximations, in many cases with respect to volumetric rendering theory. A naturally arising question is whether replacing these approximations with more principled volumetric rendering solutions can improve the quality of 3DGS. In this paper, we present an in-depth analysis of the various approximations and assumptions used by the original 3DGS solution. We demonstrate that, while more accurate volumetric rendering can help for low numbers of primitives, the power of efficient optimization and the large number of Gaussians allows 3DGS to outperform volumetric rendering despite its approximations.Item VRSurf: Surface Creation from Sparse, Unoriented 3D Strokes(The Eurographics Association and John Wiley & Sons Ltd., 2025) Sureshkumar, Anandhu; Parakkat, Amal Dev; Bonneau, Georges-Pierre; Hahmann, Stefanie; Cani, Marie-Paule; Bousseau, Adrien; Day, AngelaAlthough intuitive, sketching a closed 3D shape directly in an immersive environment results in an unordered set of arbitrary strokes, which can be difficult to assemble into a closed surface. We tackle this challenge by introducing VRSurf, a surfacing method inspired by a balloon inflation metaphor: Seeded in the sparse scaffold formed by the strokes, a smooth, closed surface is inflated to progressively interpolate the input strokes, sampled into lists of points. These are treated in a divide-and-conquer manner, which allows for automatically triggering some additional balloon inflation followed by fusion if the current inflation stops due to a detected concavity. While the input strokes are intended to belong to the same smooth 3D shape, our method is robust to coarse VR input and does not require strokes to be aligned. We simply avoid intersecting strokes that might give an inconsistent surface position due to the roughness of the VR drawing. Moreover, no additional topological information is required, and all the user needs to do is specify the initial seeding location for the first balloon. The results show that VRsurf can efficiently generate smooth surfaces that interpolate sparse sets of unoriented strokes. Validation includes a side-by-side comparison with other reconstruction methods on the same input VR sketch. We also check that our solution matches the user's intent by applying it to strokes that were sketched on an existing 3D shape and comparing what we get to the original one.Item ReConForM: Real-time Contact-aware Motion Retargeting for more Diverse Character Morphologies(The Eurographics Association and John Wiley & Sons Ltd., 2025) Cheynel, Théo; Rossi, Thomas; Bellot-Gurlet, Baptiste; Rohmer, Damien; Cani, Marie-Paule; Bousseau, Adrien; Day, AngelaPreserving semantics, in particular in terms of contacts, is a key challenge when retargeting motion between characters of different morphologies. Our solution relies on a low-dimensional embedding of the character's mesh, based on rigged key vertices that are automatically transferred from the source to the target. Motion descriptors are extracted from the trajectories of these key vertices, providing an embedding that contains combined semantic information about both shape and pose. A novel, adaptive algorithm is then used to automatically select and weight the most relevant features over time, enabling us to efficiently optimize the target motion until it conforms to these constraints, so as to preserve the semantics of the source motion. Our solution allows extensions to several novel use-cases where morphology and mesh contacts were previously overlooked, such as multi-character retargeting and motion transfer on uneven terrains. As our results show, our method is able to achieve real-time retargeting onto a wide variety of characters. Extensive experiments and comparison with state-of-the-art methods using several relevant metrics demonstrate improved results, both in terms of motion smoothness and contact accuracy.Item Neural Film Grain Rendering(The Eurographics Association and John Wiley & Sons Ltd., 2025) Lesné, Gwilherm; Gousseau, Yann; Ladjal, Saïd; Newson, Alasdair; Bousseau, Adrien; Day, AngelaFilm grain refers to the specific texture of film-acquired images, due to the physical nature of photographic film. Being a visual signature of such images, there is a strong interest in the film-industry for the rendering of these textures for digital images. Some previous works are able to closely mimic the physics of films and produce high quality results, but are computationally expensive. We propose a method based on a lightweight neural network and a texture aware loss function, achieving realistic results with very low complexity, even for large grains and high resolutions. We evaluate our algorithm both quantitatively and qualitatively with respect to previous work.