Graphical Models最新文献_第4页

Sparse support path generation for multi-axis curved layer fused filament fabrication 多轴弯曲层熔丝制造的稀疏支撑路径生成

IF 2.5 4区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Graphical Models

Pub Date : 2025-08-01 Epub Date: 2025-07-11 DOI: 10.1016/j.gmod.2025.101280

Tak Yu Lau , Dong He , Yamin Li , Yihe Wang , Danjie Bi , Lulu Huang , Pengcheng Hu , Kai Tang

In recent years, multi-axis fused filament fabrication has emerged as a solution to address the limitations of the conventional 2.5D printing process. By using a curved layering strategy and varying the print direction, the final parts can be printed with reduced support structures, enhanced surface quality, and improved mechanical properties. However, support structures in the multi-axis scheme are still needed sometimes when the support-free requirement conflicts with other constraints. Currently, most support generation algorithms are for the conventional 2.5D printing, which are not applicable to multi-axis printing. To address this issue, we propose a sparse and curved support filling pattern for multi-axis printing, aiming at enhancing the material efficiency by fully utilizing the bridge technique. Firstly, the overhang regions are detected by identifying the overhang points given a multi-axis nozzle path. Then, an optimization framework for the support guide curve is proposed to minimize its total length while ensuring that overhang filaments can be stably supported. Lastly, the support layer slices and support segments that satisfy the self-supported criterion are generated for the final support printing paths. Simulation and experiments have been performed to validate the proposed methodology.

近年来，多轴熔丝制造已经成为解决传统2.5D打印工艺局限性的一种解决方案。通过使用弯曲的分层策略和改变打印方向，最终零件可以打印出更少的支撑结构，增强表面质量，改善机械性能。然而，当无支撑要求与其他约束条件发生冲突时，仍需要在多轴方案中使用支撑结构。目前，大多数支撑生成算法都是针对传统的2.5D打印，并不适用于多轴打印。为了解决这一问题，我们提出了一种稀疏弯曲的多轴打印支撑填充图案，旨在充分利用桥接技术提高材料效率。首先，在给定多轴喷管路径的情况下，通过识别喷管的悬垂点来检测喷管的悬垂区域；然后，在保证悬垂细丝稳定支撑的前提下，提出了支撑导向曲线的优化框架，使其总长度最小。最后，生成满足自支撑条件的支撑层切片和支撑段，形成最终的支撑打印路径。仿真和实验验证了所提出的方法。

{"title":"Sparse support path generation for multi-axis curved layer fused filament fabrication","authors":"Tak Yu Lau , Dong He , Yamin Li , Yihe Wang , Danjie Bi , Lulu Huang , Pengcheng Hu , Kai Tang","doi":"10.1016/j.gmod.2025.101280","DOIUrl":"10.1016/j.gmod.2025.101280","url":null,"abstract":"<div><div>In recent years, multi-axis fused filament fabrication has emerged as a solution to address the limitations of the conventional 2.5D printing process. By using a curved layering strategy and varying the print direction, the final parts can be printed with reduced support structures, enhanced surface quality, and improved mechanical properties. However, support structures in the multi-axis scheme are still needed sometimes when the support-free requirement conflicts with other constraints. Currently, most support generation algorithms are for the conventional 2.5D printing, which are not applicable to multi-axis printing. To address this issue, we propose a sparse and curved support filling pattern for multi-axis printing, aiming at enhancing the material efficiency by fully utilizing the bridge technique. Firstly, the overhang regions are detected by identifying the overhang points given a multi-axis nozzle path. Then, an optimization framework for the support guide curve is proposed to minimize its total length while ensuring that overhang filaments can be stably supported. Lastly, the support layer slices and support segments that satisfy the self-supported criterion are generated for the final support printing paths. Simulation and experiments have been performed to validate the proposed methodology.</div></div>","PeriodicalId":55083,"journal":{"name":"Graphical Models","volume":"140 ","pages":"Article 101280"},"PeriodicalIF":2.5,"publicationDate":"2025-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144595734","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

VolumeDiffusion: Feed-forward text-to-3D generation with efficient volumetric encoder VolumeDiffusion：前馈文本到3d生成与高效的体积编码器

IF 2.5 4区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Graphical Models

Pub Date : 2025-08-01 Epub Date: 2025-06-18 DOI: 10.1016/j.gmod.2025.101274

Zhicong Tang , Shuyang Gu , Chunyu Wang , Ting Zhang , Jianmin Bao , Dong Chen , Baining Guo

This work presents VolumeDiffusion, a novel feed-forward text-to-3D generation framework that directly synthesizes 3D objects from textual descriptions. It bypasses the conventional score distillation loss based or text-to-image-to-3D approaches. To scale up the training data for the diffusion model, a novel 3D volumetric encoder is developed to efficiently acquire feature volumes from multi-view images. The 3D volumes are then trained on a diffusion model for text-to-3D generation using a 3D U-Net. This research further addresses the challenges of inaccurate object captions and high-dimensional feature volumes. The proposed model, trained on the public Objaverse dataset, demonstrates promising outcomes in producing diverse and recognizable samples from text prompts. Notably, it empowers finer control over object part characteristics through textual cues, fostering model creativity by seamlessly combining multiple concepts within a single object. This research significantly contributes to the progress of 3D generation by introducing an efficient, flexible, and scalable representation methodology.

这项工作提出了VolumeDiffusion，一个新的前馈文本到3D生成框架，直接从文本描述合成3D对象。它绕过了传统的基于分数蒸馏损失或文本到图像到3d的方法。为了扩大扩散模型的训练数据，开发了一种新的三维体积编码器，以有效地从多视图图像中获取特征体积。然后使用3D U-Net在文本到3D生成的扩散模型上训练3D体。该研究进一步解决了不准确的目标标题和高维特征量的挑战。该模型在公共Objaverse数据集上进行了训练，在从文本提示生成多样化和可识别的样本方面展示了有希望的结果。值得注意的是，它可以通过文本线索更好地控制对象部分特征，通过在单个对象中无缝地组合多个概念来培养模型创造力。该研究通过引入一种高效、灵活、可扩展的表示方法，对三维生成的进展做出了重大贡献。

{"title":"VolumeDiffusion: Feed-forward text-to-3D generation with efficient volumetric encoder","authors":"Zhicong Tang , Shuyang Gu , Chunyu Wang , Ting Zhang , Jianmin Bao , Dong Chen , Baining Guo","doi":"10.1016/j.gmod.2025.101274","DOIUrl":"10.1016/j.gmod.2025.101274","url":null,"abstract":"<div><div>This work presents VolumeDiffusion, a novel feed-forward text-to-3D generation framework that directly synthesizes 3D objects from textual descriptions. It bypasses the conventional score distillation loss based or text-to-image-to-3D approaches. To scale up the training data for the diffusion model, a novel 3D volumetric encoder is developed to efficiently acquire feature volumes from multi-view images. The 3D volumes are then trained on a diffusion model for text-to-3D generation using a 3D U-Net. This research further addresses the challenges of inaccurate object captions and high-dimensional feature volumes. The proposed model, trained on the public Objaverse dataset, demonstrates promising outcomes in producing diverse and recognizable samples from text prompts. Notably, it empowers finer control over object part characteristics through textual cues, fostering model creativity by seamlessly combining multiple concepts within a single object. This research significantly contributes to the progress of 3D generation by introducing an efficient, flexible, and scalable representation methodology.</div></div>","PeriodicalId":55083,"journal":{"name":"Graphical Models","volume":"140 ","pages":"Article 101274"},"PeriodicalIF":2.5,"publicationDate":"2025-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144314598","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Improving the area-preserving parameterization of rational Bézier surfaces by rational bilinear transformation 利用有理双线性变换改进有理bsamzier曲面的保面积参数化

IF 2.5 4区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Graphical Models

Pub Date : 2025-08-01 Epub Date: 2025-07-03 DOI: 10.1016/j.gmod.2025.101278

Xiaowei Li , Yingjie Wu , Yaohui Sun , Xin Chen , Yanru Chen , Yi-jun Yang

To improve the area-preserving parameterization quality of rational Bézier surfaces, an optimization algorithm using bilinear reparameterization is proposed. First, the rational Bézier surface is transformed using a rational bilinear transformation, which provides greater degrees of freedom compared to Möbius transformations, while preserving the rational Bézier representation. Then, the energy function is discretized using the composite Simpson’s rule, and its gradients are computed for optimization. Finally, the optimal rational bilinear transformation is determined using the L-BFGS method. Experimental results are presented to demonstrate the reparameterization effects through the circle-packing texture map, iso-parametric curve net, and color-coded images of APP energy in the proposed approach.

为了提高有理bsamzier曲面的保面积参数化质量，提出了一种双线性再参数化优化算法。首先，使用一个有理双线性变换变换有理b逍遥曲面，与Möbius变换相比，它提供了更大的自由度，同时保留了有理b逍遥表示。然后，利用复合辛普森规则对能量函数进行离散化，并计算其梯度进行优化。最后，利用L-BFGS方法确定了最优有理双线性变换。实验结果通过圆填充纹理图、等参数曲线网和APP能量彩色编码图像验证了该方法的再参数化效果。

引用次数: 0

Point cloud geometry compression based on the combination of interlayer residual and IRN concatenated residual 基于层间残差与IRN级联残差结合的点云几何压缩

IF 2.5 4区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Graphical Models

Pub Date : 2025-08-01 Epub Date: 2025-06-27 DOI: 10.1016/j.gmod.2025.101279

Meng Huang, Qian Xu, Wenxuan Xu

Point clouds have been attracting more and more attentions due to its capability of representing objects precisely, such as autonomous vehicle navigation, VR/AR, cultural heritage protection, etc. However, the enormous amount of data carried in point clouds presents significant challenges for transmission and storage. To solve this problem, this dissertation presents a point cloud compression framework based on the combination of interlayer residual and IRN concatenated residual. This paper deployed upsampling design after downsampled point cloud data. It calculates the residuals among point cloud data through downsampling and upsampling processes, consequently, maintains accuracy and reduces errors within the downsampling process. In addition, a novel Inception ResNet-Concatenated Residual Module is designed for maintaining the spatial correlation between layers and blocks. At the same time, it can extract the global and detailed features within point cloud data. Besides, Attention Module is dedicated to enhance the focus on salient features. Respectively compared with the traditional (G-PCC) and the learning point cloud compression method (PCGC v2), this paper lists a series of solid experiments data proving a 70% to 90% and a 6% to 9% BD-Rate gains on 8iVFB and Owlii datasets.

点云由于其精确表征物体的能力，在自动驾驶汽车导航、VR/AR、文化遗产保护等领域受到越来越多的关注。然而，点云中携带的大量数据对传输和存储提出了重大挑战。为了解决这一问题，本文提出了一种基于层间残差和IRN拼接残差相结合的点云压缩框架。本文对点云数据进行下采样后的上采样设计。它通过下采样和上采样计算点云数据之间的残差，从而在下采样过程中保持精度并减少误差。此外，设计了一种新颖的Inception resnet - concatated残差模块，用于保持层和块之间的空间相关性。同时，它可以提取点云数据中的全局和细节特征。此外，注意力模块致力于增强对显著特征的关注。与传统的（G-PCC）和学习点云压缩方法（PCGC v2）相比，本文列举了一系列可靠的实验数据，证明在8iVFB和Owlii数据集上的BD-Rate分别提高了70% ~ 90%和6% ~ 9%。

{"title":"Point cloud geometry compression based on the combination of interlayer residual and IRN concatenated residual","authors":"Meng Huang, Qian Xu, Wenxuan Xu","doi":"10.1016/j.gmod.2025.101279","DOIUrl":"10.1016/j.gmod.2025.101279","url":null,"abstract":"<div><div>Point clouds have been attracting more and more attentions due to its capability of representing objects precisely, such as autonomous vehicle navigation, VR/AR, cultural heritage protection, etc. However, the enormous amount of data carried in point clouds presents significant challenges for transmission and storage. To solve this problem, this dissertation presents a point cloud compression framework based on the combination of interlayer residual and IRN concatenated residual. This paper deployed upsampling design after downsampled point cloud data. It calculates the residuals among point cloud data through downsampling and upsampling processes, consequently, maintains accuracy and reduces errors within the downsampling process. In addition, a novel Inception ResNet-Concatenated Residual Module is designed for maintaining the spatial correlation between layers and blocks. At the same time, it can extract the global and detailed features within point cloud data. Besides, Attention Module is dedicated to enhance the focus on salient features. Respectively compared with the traditional (G-PCC) and the learning point cloud compression method (PCGC v2), this paper lists a series of solid experiments data proving a 70% to 90% and a 6% to 9% BD-Rate gains on 8iVFB and Owlii datasets.</div></div>","PeriodicalId":55083,"journal":{"name":"Graphical Models","volume":"140 ","pages":"Article 101279"},"PeriodicalIF":2.5,"publicationDate":"2025-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144490700","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

RS-SpecSDF: Reflection-supervised surface reconstruction and material estimation for specular indoor scenes RS-SpecSDF：镜面室内场景的反射监督表面重建和材料估计

IF 2.5 4区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Graphical Models

Pub Date : 2025-08-01 Epub Date: 2025-06-25 DOI: 10.1016/j.gmod.2025.101277

Dong-Yu Chen, Hao-Xiang Chen, Qun-Ce Xu, Tai-Jiang Mu

Neural Radiance Field (NeRF) has achieved impressive 3D reconstruction quality using implicit scene representations. However, planar specular reflections pose significant challenges in the 3D reconstruction task. It is a common practice to decompose the scene into physically real geometries and virtual images produced by the reflections. However, current methods struggle to resolve the ambiguities in the decomposition process, because they mostly rely on mirror masks as external cues. They also fail to acquire accurate surface materials, which is essential for downstream applications of the recovered geometries. In this paper, we present RS-SpecSDF, a novel framework for indoor scene surface reconstruction that can faithfully reconstruct specular reflectors while accurately decomposing the reflection from the scene geometries and recovering the accurate specular fraction and diffuse appearance of the surface without requiring mirror masks. Our key idea is to perform reflection ray-casting and use it as supervision for the decomposition of reflection and surface material. Our method is based on an observation that the virtual image seen by the camera ray should be consistent with the object that the ray hits after reflecting off the specular surface. To leverage this constraint, we propose the Reflection Consistency Loss and Reflection Certainty Loss to regularize the decomposition. Experiments conducted on both our newly-proposed synthetic dataset and a real-captured dataset demonstrate that our method achieves high-quality surface reconstruction and accurate material decomposition results without the need of mirror masks.

神经辐射场（NeRF）已经实现了令人印象深刻的3D重建质量使用隐式场景表示。然而，平面镜面反射在三维重建任务中提出了重大挑战。将场景分解为物理上真实的几何图形和由反射产生的虚拟图像是一种常见的做法。然而，目前的方法很难解决分解过程中的模糊性，因为它们主要依赖于镜像掩模作为外部线索。它们也无法获得精确的表面材料，这对于回收几何形状的下游应用至关重要。在本文中，我们提出了一种新的室内场景表面重建框架RS-SpecSDF，它可以忠实地重建镜面反射器，同时准确地从场景几何形状中分解反射，并在不需要镜面掩模的情况下恢复表面的精确镜面分数和漫反射外观。我们的主要想法是执行反射光线投射，并使用它作为反射和表面材料分解的监督。我们的方法是基于这样一种观察，即相机光线所看到的虚拟图像应该与光线从镜面反射后击中的物体一致。为了利用这一约束，我们提出了反射一致性损失和反射确定性损失来正则化分解。在合成数据集和实际捕获数据集上进行的实验表明，我们的方法在不需要镜像掩模的情况下获得了高质量的表面重建和准确的材料分解结果。

{"title":"RS-SpecSDF: Reflection-supervised surface reconstruction and material estimation for specular indoor scenes","authors":"Dong-Yu Chen, Hao-Xiang Chen, Qun-Ce Xu, Tai-Jiang Mu","doi":"10.1016/j.gmod.2025.101277","DOIUrl":"10.1016/j.gmod.2025.101277","url":null,"abstract":"<div><div>Neural Radiance Field (NeRF) has achieved impressive 3D reconstruction quality using implicit scene representations. However, planar specular reflections pose significant challenges in the 3D reconstruction task. It is a common practice to decompose the scene into physically real geometries and virtual images produced by the reflections. However, current methods struggle to resolve the ambiguities in the decomposition process, because they mostly rely on mirror masks as external cues. They also fail to acquire accurate surface materials, which is essential for downstream applications of the recovered geometries. In this paper, we present RS-SpecSDF, a novel framework for indoor scene surface reconstruction that can faithfully reconstruct specular reflectors while accurately decomposing the reflection from the scene geometries and recovering the accurate specular fraction and diffuse appearance of the surface without requiring mirror masks. Our key idea is to perform reflection ray-casting and use it as supervision for the decomposition of reflection and surface material. Our method is based on an observation that the virtual image seen by the camera ray should be consistent with the object that the ray hits after reflecting off the specular surface. To leverage this constraint, we propose the Reflection Consistency Loss and Reflection Certainty Loss to regularize the decomposition. Experiments conducted on both our newly-proposed synthetic dataset and a real-captured dataset demonstrate that our method achieves high-quality surface reconstruction and accurate material decomposition results without the need of mirror masks.</div></div>","PeriodicalId":55083,"journal":{"name":"Graphical Models","volume":"140 ","pages":"Article 101277"},"PeriodicalIF":2.5,"publicationDate":"2025-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144472382","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

L2-GNN: Graph neural networks with fast spectral filters using twice linear parameterization L2-GNN：使用两次线性参数化的快速光谱滤波的图神经网络

IF 2.5 4区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Graphical Models

Pub Date : 2025-08-01 Epub Date: 2025-06-26 DOI: 10.1016/j.gmod.2025.101276

Siying Huang , Xin Yang , Zhengda Lu , Hongxing Qin , Huaiwen Zhang , Yiqun Wang

To improve learning on irregular 3D shapes, such as meshes with varying discretizations and point clouds with different samplings, we propose L

^{2}

-GNN, a new graph neural network that approximates the spectral filters using twice linear parameterization. First, we parameterize the spectral filters using wavelet filter basis functions. The parameterization allows for an enlarged receptive field of graph convolutions, which can simultaneously capture low-frequency and high-frequency information. Second, we parameterize the wavelet filter basis functions using Chebyshev polynomial basis functions. This parameterization reduces the computational complexity of graph convolutions while maintaining robustness to the change of mesh discretization and point cloud sampling. Our L

^{2}

-GNN based on the fast spectral filter can be used for shape correspondence, classification, and segmentation tasks on non-regular mesh or point cloud data. Experimental results show that our method outperforms the current state of the art in terms of both quality and efficiency.

为了提高不规则三维形状的学习能力，例如具有不同离散化的网格和不同采样的点云，我们提出了L2-GNN，一种新的图神经网络，它使用两次线性参数化来近似光谱滤波器。首先，利用小波滤波基函数对谱滤波器进行参数化。参数化允许扩大图卷积的接受域，可以同时捕获低频和高频信息。其次，利用切比雪夫多项式基函数对小波滤波器基函数进行参数化。这种参数化降低了图卷积的计算复杂度，同时保持了对网格离散化和点云采样变化的鲁棒性。基于快速光谱滤波的L2-GNN可用于不规则网格或点云数据的形状对应、分类和分割任务。实验结果表明，我们的方法在质量和效率方面都优于目前的技术水平。

{"title":"L2-GNN: Graph neural networks with fast spectral filters using twice linear parameterization","authors":"Siying Huang , Xin Yang , Zhengda Lu , Hongxing Qin , Huaiwen Zhang , Yiqun Wang","doi":"10.1016/j.gmod.2025.101276","DOIUrl":"10.1016/j.gmod.2025.101276","url":null,"abstract":"<div><div>To improve learning on irregular 3D shapes, such as meshes with varying discretizations and point clouds with different samplings, we propose L<span><math><msup><mrow></mrow><mrow><mn>2</mn></mrow></msup></math></span>-GNN, a new graph neural network that approximates the spectral filters using twice linear parameterization. First, we parameterize the spectral filters using wavelet filter basis functions. The parameterization allows for an enlarged receptive field of graph convolutions, which can simultaneously capture low-frequency and high-frequency information. Second, we parameterize the wavelet filter basis functions using Chebyshev polynomial basis functions. This parameterization reduces the computational complexity of graph convolutions while maintaining robustness to the change of mesh discretization and point cloud sampling. Our L<span><math><msup><mrow></mrow><mrow><mn>2</mn></mrow></msup></math></span>-GNN based on the fast spectral filter can be used for shape correspondence, classification, and segmentation tasks on non-regular mesh or point cloud data. Experimental results show that our method outperforms the current state of the art in terms of both quality and efficiency.</div></div>","PeriodicalId":55083,"journal":{"name":"Graphical Models","volume":"140 ","pages":"Article 101276"},"PeriodicalIF":2.5,"publicationDate":"2025-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144490699","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

LDM: Large tensorial SDF model for textured mesh generation LDM：用于纹理网格生成的大张量SDF模型

IF 2.5 4区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Graphical Models

Pub Date : 2025-08-01 Epub Date: 2025-06-21 DOI: 10.1016/j.gmod.2025.101271

Rengan Xie , Kai Huang , Xiaoliang Luo , Yizheng Chen , Lvchun Wang , Qi Wang , Qi Ye , Wei Chen , Wenting Zheng , Yuchi Huo

Previous efforts have managed to generate production-ready 3D assets from text or images. However, these methods primarily employ NeRF or 3D Gaussian representations, which are not adept at producing smooth, high-quality geometries required by modern rendering pipelines. In this paper, we propose LDM, a Large tensorial SDF Model, which introduces a novel feed-forward framework capable of generating high-fidelity, illumination-decoupled textured mesh from a single image or text prompts. We firstly utilize a multi-view diffusion model to generate sparse multi-view inputs from single images or text prompts, and then a transformer-based model is trained to predict a tensorial SDF field from these sparse multi-view image inputs. Finally, we employ a gradient-based mesh optimization layer to refine this model, enabling it to produce an SDF field from which high-quality textured meshes can be extracted. Extensive experiments demonstrate that our method can generate diverse, high-quality 3D mesh assets with corresponding decomposed RGB textures within seconds. The project code is available at https://github.com/rgxie/LDM.

之前的努力已经成功地从文本或图像生成生产就绪的3D资产。然而，这些方法主要采用NeRF或3D高斯表示，它们不擅长生成现代渲染管道所需的光滑、高质量的几何形状。在本文中，我们提出了LDM，一种大型张量SDF模型，它引入了一种新的前馈框架，能够从单个图像或文本提示生成高保真度，光照解耦的纹理网格。我们首先利用多视图扩散模型从单个图像或文本提示生成稀疏多视图输入，然后训练基于变换的模型从这些稀疏多视图图像输入预测张量SDF场。最后，我们使用基于梯度的网格优化层来细化该模型，使其能够产生一个SDF场，从中可以提取高质量的纹理网格。大量的实验表明，我们的方法可以在几秒钟内生成具有相应分解RGB纹理的多种高质量3D网格资产。项目代码可从https://github.com/rgxie/LDM获得。

{"title":"LDM: Large tensorial SDF model for textured mesh generation","authors":"Rengan Xie , Kai Huang , Xiaoliang Luo , Yizheng Chen , Lvchun Wang , Qi Wang , Qi Ye , Wei Chen , Wenting Zheng , Yuchi Huo","doi":"10.1016/j.gmod.2025.101271","DOIUrl":"10.1016/j.gmod.2025.101271","url":null,"abstract":"<div><div>Previous efforts have managed to generate production-ready 3D assets from text or images. However, these methods primarily employ NeRF or 3D Gaussian representations, which are not adept at producing smooth, high-quality geometries required by modern rendering pipelines. In this paper, we propose LDM, a <strong>L</strong>arge tensorial S<strong>D</strong>F <strong>M</strong>odel, which introduces a novel feed-forward framework capable of generating high-fidelity, illumination-decoupled textured mesh from a single image or text prompts. We firstly utilize a multi-view diffusion model to generate sparse multi-view inputs from single images or text prompts, and then a transformer-based model is trained to predict a tensorial SDF field from these sparse multi-view image inputs. Finally, we employ a gradient-based mesh optimization layer to refine this model, enabling it to produce an SDF field from which high-quality textured meshes can be extracted. Extensive experiments demonstrate that our method can generate diverse, high-quality 3D mesh assets with corresponding decomposed RGB textures within seconds. The project code is available at <span><span>https://github.com/rgxie/LDM</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":55083,"journal":{"name":"Graphical Models","volume":"140 ","pages":"Article 101271"},"PeriodicalIF":2.5,"publicationDate":"2025-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144330266","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

DDD++: Exploiting Density map consistency for Deep Depth estimation in indoor environments dddd++：利用密度图一致性在室内环境中进行深度估计

IF 2.5 4区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Graphical Models

Pub Date : 2025-08-01 Epub Date: 2025-07-22 DOI: 10.1016/j.gmod.2025.101281

Giovanni Pintore , Marco Agus , Alberto Signoroni , Enrico Gobbetti

We introduce a novel deep neural network designed for fast and structurally consistent monocular 360° depth estimation in indoor settings. Our model generates a spherical depth map from a single gravity-aligned or gravity-rectified equirectangular image, ensuring the predicted depth aligns with the typical depth distribution and structural features of cluttered indoor spaces, which are generally enclosed by walls, floors, and ceilings. By leveraging the distinctive vertical and horizontal patterns found in man-made indoor environments, we propose a streamlined network architecture that incorporates gravity-aligned feature flattening and specialized vision transformers. Through flattening, these transformers fully exploit the omnidirectional nature of the input without requiring patch segmentation or positional encoding. To further enhance structural consistency, we introduce a novel loss function that assesses density map consistency by projecting points from the predicted depth map onto a horizontal plane and a cylindrical proxy. This lightweight architecture requires fewer tunable parameters and computational resources than competing methods. Our comparative evaluation shows that our approach improves depth estimation accuracy while ensuring greater structural consistency compared to existing methods. For these reasons, it promises to be suitable for incorporation in real-time solutions, as well as a building block in more complex structural analysis and segmentation methods.

我们介绍了一种新的深度神经网络，用于快速和结构一致的室内单目360°深度估计。我们的模型从单个重力对齐或重力校正的等矩形图像生成球形深度图，确保预测深度与典型的深度分布和杂乱室内空间的结构特征保持一致，这些空间通常被墙壁、地板和天花板包围。通过利用人造室内环境中独特的垂直和水平模式，我们提出了一个流线型的网络架构，该架构结合了重力对齐的特征平坦化和专门的视觉变压器。通过平坦化，这些变压器充分利用了输入的全向特性，而不需要补丁分割或位置编码。为了进一步增强结构一致性，我们引入了一种新的损失函数，通过将预测深度图中的点投影到水平面和圆柱形代理上来评估密度图的一致性。与竞争方法相比，这种轻量级体系结构需要更少的可调参数和计算资源。我们的对比评估表明，与现有方法相比，我们的方法提高了深度估计精度，同时确保了更大的结构一致性。由于这些原因，它有望适用于实时解决方案的整合，以及更复杂的结构分析和分割方法的构建块。

{"title":"DDD++: Exploiting Density map consistency for Deep Depth estimation in indoor environments","authors":"Giovanni Pintore , Marco Agus , Alberto Signoroni , Enrico Gobbetti","doi":"10.1016/j.gmod.2025.101281","DOIUrl":"10.1016/j.gmod.2025.101281","url":null,"abstract":"<div><div>We introduce a novel deep neural network designed for fast and structurally consistent monocular 360° depth estimation in indoor settings. Our model generates a spherical depth map from a single gravity-aligned or gravity-rectified equirectangular image, ensuring the predicted depth aligns with the typical depth distribution and structural features of cluttered indoor spaces, which are generally enclosed by walls, floors, and ceilings. By leveraging the distinctive vertical and horizontal patterns found in man-made indoor environments, we propose a streamlined network architecture that incorporates gravity-aligned feature flattening and specialized vision transformers. Through flattening, these transformers fully exploit the omnidirectional nature of the input without requiring patch segmentation or positional encoding. To further enhance structural consistency, we introduce a novel loss function that assesses density map consistency by projecting points from the predicted depth map onto a horizontal plane and a cylindrical proxy. This lightweight architecture requires fewer tunable parameters and computational resources than competing methods. Our comparative evaluation shows that our approach improves depth estimation accuracy while ensuring greater structural consistency compared to existing methods. For these reasons, it promises to be suitable for incorporation in real-time solutions, as well as a building block in more complex structural analysis and segmentation methods.</div></div>","PeriodicalId":55083,"journal":{"name":"Graphical Models","volume":"140 ","pages":"Article 101281"},"PeriodicalIF":2.5,"publicationDate":"2025-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144679962","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

DC-APIC: A decomposed compatible affine particle in cell transfer scheme for non-sticky solid–fluid interactions in MPM DC-APIC：在MPM中非粘性固-液相互作用的细胞转移方案中分解的兼容仿射粒子

IF 2.5 4区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Graphical Models

Pub Date : 2025-06-01 Epub Date: 2025-05-25 DOI: 10.1016/j.gmod.2025.101269

Chenhui Wang , Jianyang Zhang , Chen Li , Changbo Wang

Despite the material point method (MPM) provides a unified particle simulation framework for coupling of different materials, MPM suffers from sticky numerical artifacts, which is inherently restricted to sticky and no-slip interactions. In this paper, we propose a novel transfer scheme called Decomposed Compatible Affine Particle in Cell (DC-APIC) within the MPM framework for simulating the two-way coupled interaction between elastic solids and incompressible fluids under free-slip boundary conditions on a unified background grid. Firstly, we adopt particle-grid compatibility to describe the relationship between grid nodes and particles at the fluid–solid interface, which serves as the guideline for subsequent particle–grid–particle transfers. Then we develop a phase-field gradient method to track the compatibility and normal directions at the interface. Secondly, to facilitate automatic MPM collision resolution during solid–fluid coupling, in the proposed DC-APIC integrator, the tangential component will not be transferred between incompatible grid nodes to prevent velocity smoothing in another phase, while the normal component is transferred without limitations. Finally, our comprehensive results confirm that our approach effectively reduces diffusion and unphysical viscosity compared to traditional MPM.

尽管材料点法（MPM）为不同材料的耦合提供了统一的粒子模拟框架，但MPM存在粘性数值伪影，固有地局限于粘性和无滑移相互作用。本文在MPM框架下，提出了一种新的可分解兼容仿射粒子单元（DC-APIC）传输方案，用于模拟统一背景网格下自由滑移边界条件下弹性固体与不可压缩流体之间的双向耦合相互作用。首先，我们采用颗粒-网格相容性来描述流固界面上网格节点与颗粒之间的关系，为后续颗粒-网格-颗粒转移提供指导。然后，我们提出了一种相场梯度法来跟踪界面处的相容性和法线方向。其次，为了方便固流耦合过程中MPM碰撞自动解决，在本文提出的DC-APIC积分器中，切向分量不会在不兼容的网格节点之间转移，以防止另一阶段的速度平滑，而法向分量的转移没有限制。最后，我们的综合结果证实，与传统的MPM相比，我们的方法有效地降低了扩散和非物理粘度。

{"title":"DC-APIC: A decomposed compatible affine particle in cell transfer scheme for non-sticky solid–fluid interactions in MPM","authors":"Chenhui Wang , Jianyang Zhang , Chen Li , Changbo Wang","doi":"10.1016/j.gmod.2025.101269","DOIUrl":"10.1016/j.gmod.2025.101269","url":null,"abstract":"<div><div>Despite the material point method (MPM) provides a unified particle simulation framework for coupling of different materials, MPM suffers from sticky numerical artifacts, which is inherently restricted to sticky and no-slip interactions. In this paper, we propose a novel transfer scheme called Decomposed Compatible Affine Particle in Cell (DC-APIC) within the MPM framework for simulating the two-way coupled interaction between elastic solids and incompressible fluids under free-slip boundary conditions on a unified background grid. Firstly, we adopt particle-grid compatibility to describe the relationship between grid nodes and particles at the fluid–solid interface, which serves as the guideline for subsequent particle–grid–particle transfers. Then we develop a phase-field gradient method to track the compatibility and normal directions at the interface. Secondly, to facilitate automatic MPM collision resolution during solid–fluid coupling, in the proposed DC-APIC integrator, the tangential component will not be transferred between incompatible grid nodes to prevent velocity smoothing in another phase, while the normal component is transferred without limitations. Finally, our comprehensive results confirm that our approach effectively reduces diffusion and unphysical viscosity compared to traditional MPM.</div></div>","PeriodicalId":55083,"journal":{"name":"Graphical Models","volume":"139 ","pages":"Article 101269"},"PeriodicalIF":2.5,"publicationDate":"2025-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144134591","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Semantics-aware human motion generation from audio instructions 从音频指令生成语义感知的人体运动

IF 2.5 4区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Graphical Models

Pub Date : 2025-06-01 Epub Date: 2025-05-23 DOI: 10.1016/j.gmod.2025.101268

Zi-An Wang , Shihao Zou , Shiyao Yu , Mingyuan Zhang , Chao Dong

Recent advances in interactive technologies have highlighted the prominence of audio signals for semantic encoding. This paper explores a new task, where audio signals are used as conditioning inputs to generate motions that align with the semantics of the audio. Unlike text-based interactions, audio provides a more natural and intuitive communication method. However, existing methods typically focus on matching motions with music or speech rhythms, which often results in a weak connection between the semantics of the audio and generated motions. We propose an end-to-end framework using a masked generative transformer, enhanced by a memory-retrieval attention module to handle sparse and lengthy audio inputs. Additionally, we enrich existing datasets by converting descriptions into conversational style and generating corresponding audio with varied speaker identities. Experiments demonstrate the effectiveness and efficiency of the proposed framework, demonstrating that audio instructions can convey semantics similar to text while providing more practical and user-friendly interactions.

交互技术的最新进展突出了音频信号在语义编码中的重要性。本文探索了一个新的任务，其中音频信号被用作条件反射输入，以产生与音频语义一致的运动。与基于文本的交互不同，音频提供了一种更自然、更直观的交流方式。然而，现有的方法通常侧重于将动作与音乐或语音节奏相匹配，这往往导致音频的语义与生成的动作之间的联系很弱。我们提出了一个端到端框架，使用屏蔽生成转换器，增强了一个记忆检索注意力模块来处理稀疏和冗长的音频输入。此外，我们通过将描述转换为会话风格并生成具有不同说话者身份的相应音频来丰富现有数据集。实验证明了该框架的有效性和效率，表明音频指令可以传达与文本相似的语义，同时提供更实用和用户友好的交互。

引用次数: 0