Search

Unsigned distance fields (UDFs) allow for the representation of models with complex topologies, but extracting accurate zero level sets from these fields poses significant challenges, particularly in preserving topological accuracy and capturing fine geometric details. To overcome these issues, we introduce DCUDF2, an enhancement over DCUDF--the current state-of-the-art method--for extracting zero level sets from UDFs. Our approach utilizes an accuracy-aware loss function, enhanced with self-adaptive weights, to improve geometric quality significantly. We also propose a topology correction strategy that reduces the dependence on hyper-parameter, increasing the robustness of our method. Furthermore, we develop new operations leveraging self-adaptive weights to boost runtime efficiency. Extensive experiments on surface extraction across diverse datasets demonstrate that DCUDF2 outperforms DCUDF and existing methods in both geometric fidelity and topological accuracy. We will make the source code publicly available.

Numerical solvers of Partial Differential Equations (PDEs) are of fundamental significance to science and engineering. To date, the historical reliance on legacy techniques has circumscribed possible integration of big data knowledge and exhibits sub-optimal efficiency for certain PDE formulations, while data-driven neural methods typically lack mathematical guarantee of convergence and correctness. This paper articulates a mathematically rigorous neural solver for linear PDEs. The proposed UGrid solver, built upon the principled integration of U-Net and MultiGrid, manifests a mathematically rigorous proof of both convergence and correctness, and showcases high numerical accuracy, as well as strong generalization power to various input geometry/values and multiple PDE formulations. In addition, we devise a new residual loss metric, which enables unsupervised training and affords more stability and a larger solution space over the legacy losses., Comment: Proceedings of the 41st International Conference on Machine Learning, Vienna, Austria. PMLR 235, 2024

Unsigned distance fields (UDFs) provide a versatile framework for representing a diverse array of 3D shapes, encompassing both watertight and non-watertight geometries. Traditional UDF learning methods typically require extensive training on large datasets of 3D shapes, which is costly and often necessitates hyperparameter adjustments for new datasets. This paper presents a novel neural framework, LoSF-UDF, for reconstructing surfaces from 3D point clouds by leveraging local shape functions to learn UDFs. We observe that 3D shapes manifest simple patterns within localized areas, prompting us to create a training dataset of point cloud patches characterized by mathematical functions that represent a continuum from smooth surfaces to sharp edges and corners. Our approach learns features within a specific radius around each query point and utilizes an attention mechanism to focus on the crucial features for UDF estimation. This method enables efficient and robust surface reconstruction from point clouds without the need for shape-specific training. Additionally, our method exhibits enhanced resilience to noise and outliers in point clouds compared to existing methods. We present comprehensive experiments and comparisons across various datasets, including synthetic and real-scanned point clouds, to validate our method's efficacy., Comment: 14 pages, 11 figures

The 3D Gaussian Splatting technique has significantly advanced the construction of radiance fields from multi-view images, enabling real-time rendering. While point-based rasterization effectively reduces computational demands for rendering, it often struggles to accurately reconstruct the geometry of the target object, especially under strong lighting. To address this challenge, we introduce a novel approach that combines octree-based implicit surface representations with Gaussian splatting. Our method consists of four stages. Initially, it reconstructs a signed distance field (SDF) and a radiance field through volume rendering, encoding them in a low-resolution octree. The initial SDF represents the coarse geometry of the target object. Subsequently, it introduces 3D Gaussians as additional degrees of freedom, which are guided by the SDF. In the third stage, the optimized Gaussians further improve the accuracy of the SDF, allowing it to recover finer geometric details compared to the initial SDF obtained in the first stage. Finally, it adopts the refined SDF to further optimize the 3D Gaussians via splatting, eliminating those that contribute little to visual appearance. Experimental results show that our method, which leverages the distribution of 3D Gaussians with SDFs, reconstructs more accurate geometry, particularly in images with specular highlights caused by strong lighting.

While Signed Distance Fields (SDF) are well-established for modeling watertight surfaces, Unsigned Distance Fields (UDF) broaden the scope to include open surfaces and models with complex inner structures. Despite their flexibility, UDFs encounter significant challenges in high-fidelity 3D reconstruction, such as non-differentiability at the zero level set, difficulty in achieving the exact zero value, numerous local minima, vanishing gradients, and oscillating gradient directions near the zero level set. To address these challenges, we propose Details Enhanced UDF (DEUDF) learning that integrates normal alignment and the SIREN network for capturing fine geometric details, adaptively weighted Eikonal constraints to address vanishing gradients near the target surface, unconditioned MLP-based UDF representation to relax non-negativity constraints, and a UDF-tailored method for extracting iso-surface with non-constant iso-values. These strategies collectively stabilize the learning process from unoriented point clouds and enhance the accuracy of UDFs. Our computational results demonstrate that DEUDF outperforms existing UDF learning methods in both accuracy and the quality of reconstructed surfaces. We will make the source code publicly available.

This paper presents a method for reconstructing watertight 3D surfaces from unoriented point clouds. Starting with randomly initialized normals, the method iteratively refines each normal by diffusing the gradient of the generalized winding number (GWN) field. Upon convergence, the target surface is extracted using the standard Marching Cubes algorithm. Our method is conceptually simple, easy to implement, and does not require numerical solvers, which distinguishes it from existing approaches. Designed for parallelization and scalability, it efficiently handles large-scale models on both CPUs and GPUs. Experimental results demonstrate that our method outperforms all existing methods in reconstructing from unoriented point clouds, particularly in terms of runtime performance. On large-scale models with 10 to 20 million points, our CUDA implementation on an NVIDIA GTX 4090 GPU is typically 30-100x faster than iPSR, the leading sequential method tested on a high-end PC with an Intel i9 CPU. Furthermore, our approach exhibits superior robustness against noise and effectively handles models with thin structures, surpassing existing methods. We will make the source code publicly available to encourage further research and applications.

In this paper, we are concerned with the global existence and scattering of small data smooth solutions to a class of quasilinear wave systems on the product space $\mathbb{R}^2\times\mathbb{T}$. These quasilinear wave systems include 3D irrotational potential flow equation of Chaplygin gases, 3D relativistic membrane equation, some 3D quasilinear wave equations which come from the corresponding Lagrangian functionals as perturbations of the Lagrangian densities of linear waves, and nonlinear wave maps system. Through looking for some suitable transformations of unknown functions, the nonlinear wave system can be reduced into a more tractable form. Subsequently, by applying the vector-field method together with the ghost weight technique as well as deriving some kinds of weighted $L^\infty-L^\infty$ and $L^\infty-L^2$ estimates of solution $w$ to the 2D linear wave equation $\Box w=f(t,x)$, the global existence and scattering of small data solutions are established.

We introduce a new generative model that combines latent diffusion with persistent homology to create 3D shapes with high diversity, with a special emphasis on their topological characteristics. Our method involves representing 3D shapes as implicit fields, then employing persistent homology to extract topological features, including Betti numbers and persistence diagrams. The shape generation process consists of two steps. Initially, we employ a transformer-based autoencoding module to embed the implicit representation of each 3D shape into a set of latent vectors. Subsequently, we navigate through the learned latent space via a diffusion model. By strategically incorporating topological features into the diffusion process, our generative module is able to produce a richer variety of 3D shapes with different topological structures. Furthermore, our framework is flexible, supporting generation tasks constrained by a variety of inputs, including sparse and partial point clouds, as well as sketches. By modifying the persistence diagrams, we can alter the topology of the shapes generated from these input modalities., Comment: 16 pages, 9 figures

The advancements in neural rendering have increased the need for techniques that enable intuitive editing of 3D objects represented as neural implicit surfaces. This paper introduces a novel neural algorithm for parameterizing neural implicit surfaces to simple parametric domains like spheres and polycubes. Our method allows users to specify the number of cubes in the parametric domain, learning a configuration that closely resembles the target 3D object's geometry. It computes bi-directional deformation between the object and the domain using a forward mapping from the object's zero level set and an inverse deformation for backward mapping. We ensure nearly bijective mapping with a cycle loss and optimize deformation smoothness. The parameterization quality, assessed by angle and area distortions, is guaranteed using a Laplacian regularizer and an optimized learned parametric domain. Our framework integrates with existing neural rendering pipelines, using multi-view images of a single object or multiple objects of similar geometries to reconstruct 3D geometry and compute texture maps automatically, eliminating the need for any prior information. We demonstrate the method's effectiveness on images of human heads and man-made objects., Comment: ECCV24, see https://xubaixinxbx.github.io/neuparam

In this paper, we propose a new method, called DoubleCoverUDF, for extracting the zero level-set from unsigned distance fields (UDFs). DoubleCoverUDF takes a learned UDF and a user-specified parameter $r$ (a small positive real number) as input and extracts an iso-surface with an iso-value $r$ using the conventional marching cubes algorithm. We show that the computed iso-surface is the boundary of the $r$-offset volume of the target zero level-set $S$, which is an orientable manifold, regardless of the topology of $S$. Next, the algorithm computes a covering map to project the boundary mesh onto $S$, preserving the mesh's topology and avoiding folding. If $S$ is an orientable manifold surface, our algorithm separates the double-layered mesh into a single layer using a robust minimum-cut post-processing step. Otherwise, it keeps the double-layered mesh as the output. We validate our algorithm by reconstructing 3D surfaces of open models and demonstrate its efficacy and effectiveness on synthetic models and benchmark datasets. Our experimental results confirm that our method is robust and produces meshes with better quality in terms of both visual evaluation and quantitative measures than existing UDF-based methods. The source code is available at https://github.com/jjjkkyz/DCUDF., Comment: published in ACM Transactions on Graphics (SIGGRAPH Asia 2023)

Chalcogenide perovskites, such as BaZrS$_3$, are emerging semiconductors with potential for high photovoltaic power conversion efficiency. The role of defects in the efficiency of the generation and collection of photo-excited carriers has not been experimentally investigated extensively. We study the effect of processing-induced defects on the photoconductive properties of single crystals of BaZrS$_3$. We achieved ohmic contacts to single crystals of BaZrS$_3$ and observed positive surface photovoltage, which is typically observed in p-type semiconductors. However, mechanical polishing of BaZrS$_3$ to remove the surface oxide leads to dense deformation grain boundaries and leads to trap-dominated photoconductive response. In comparison, ohmic contacts achieved in cleaved crystals leave fewer deformation defects and greatly improve optoelectronic properties. Defect-controlled crystal growth and contact fabrication are potentially limiting factors for achieving high photon-to-excited electron conversion efficiency in BaZrS$_3$.

It has been known that if the initial data decay sufficiently fast at space infinity, then 1D Klein-Gordon equations with quadratic nonlinearity admit classical solutions up to time $e^{C/\epsilon^2}$ while $e^{C/\epsilon^2}$ is also the upper bound of the lifespan, where $C>0$ is some suitable constant and $\epsilon>0$ is the size of the initial data. In this paper, we will focus on the 1D nonlinear Klein-Gordon equations with weakly decaying initial data. It is shown that if the $H^s$-Sobolev norm with $(1+|x|)^{1/2+}$ weight of the initial data is small, then the almost global solutions exist; if the initial $H^s$-Sobolev norm with $(1+|x|)^{1/2}$ weight is small, then for any $M>0$, the solutions exist on $[0,\epsilon^{-M}]$. Our proof is based on the dispersive estimate with a suitable $Z$-norm and a delicate analysis on the phase function.

Image smoothing is by reducing pixel-wise gradients to smooth out details. As existing methods always rely on gradients to determine smoothing manners, it is difficult to distinguish structures and details to handle distinctively due to the overlapped ranges of gradients for structures and details. Thus, it is still challenging to achieve high-quality results, especially on preserving weak structures and removing high-contrast details. In this paper, we address this challenge by improving the real-time optimization-based method via iterative least squares (called ILS). We observe that 1) ILS uses gradients as the independent variable in its penalty function for determining smoothing manners, and 2) the framework of ILS can still work for image smoothing when we use some values instead of gradients in the penalty function. Thus, corresponding to the properties of pixels on structures or not, we compute some values to use in the penalty function to determine smoothing manners, and so we can handle structures and details distinctively, no matter whether their gradients are high or low. As a result, we can conveniently remove high-contrast details while preserving weak structures. Moreover, such values can be adjusted to accelerate optimization computation, so that we can use fewer iterations than the original ILS method for efficiency. This also reduces the changes onto structures to help structure preservation. Experimental results show our advantages over existing methods on efficiency and quality.

Recently, building on the foundation of neural radiance field, various techniques have emerged to learn unsigned distance fields (UDF) to reconstruct 3D non-watertight models from multi-view images. Yet, a central challenge in UDF-based volume rendering is formulating a proper way to convert unsigned distance values into volume density, ensuring that the resulting weight function remains unbiased and sensitive to occlusions. Falling short on these requirements often results in incorrect topology or large reconstruction errors in resulting models. This paper addresses this challenge by presenting a novel two-stage algorithm, 2S-UDF, for learning a high-quality UDF from multi-view images. Initially, the method applies an easily trainable density function that, while slightly biased and transparent, aids in coarse reconstruction. The subsequent stage then refines the geometry and appearance of the object to achieve a high-quality reconstruction by directly adjusting the weight function used in volume rendering to ensure that it is unbiased and occlusion-aware. Decoupling density and weight in two stages makes our training stable and robust, distinguishing our technique from existing UDF learning approaches. Evaluations on the DeepFashion3D, DTU, and BlendedMVS datasets validate the robustness and effectiveness of our proposed approach. In both quantitative metrics and visual quality, the results indicate our superior performance over other UDF learning techniques in reconstructing 3D non-watertight models from multi-view images. Our code is available at https://bitbucket.org/jkdeng/2sudf/., Comment: accepted to CVPR 2024

In this paper, we investigate the fully nonlinear wave equations on the product space $\mathbb{R}^3\times\mathbb{T}$ with quadratic nonlinearities and on $\mathbb{R}^2\times\mathbb{T}$ with cubic nonlinearities, respectively. It is shown that for the small initial data satisfying some space-decay rates at infinity, these nonlinear equations admit global smooth solutions when the corresponding partial null conditions hold and while have almost global smooth solutions when the partial null conditions are violated. Our proof relies on the Fourier mode decomposition of the solutions with respect to the periodic direction, the efficient combinations of time-decay estimates for the solutions to the linear wave equations and the linear Klein-Gordon equations, and the global weighted energy estimates. In addition, an interesting auxiliary energy is introduced. As a byproduct, our results can be applied to the 4D irrotational compressible Euler equations of polytropic gases or Chaplygin gases on $\mathbb{R}^3\times\mathbb{T}$, the 3D relativistic membrane equation and the 3D nonlinear membrane equation on $\mathbb{R}^2\times\mathbb{T}$.

It is well known that for the quasilinear Klein-Gordon equation with quadratic nonlinearity and sufficiently decaying small initial data, there exists a global smooth solution if the space dimensions $d\geq2$. When the initial data are of size $\varepsilon>0$ in the Sobolev space, for the semilinear Klein-Gordon equation satisfying the null condition, the authors in the article (J.-M. Delort, Daoyuan Fang, Almost global existence for solutions of semilinear Klein-Gordon equations with small weakly decaying Cauchy data, Comm. Partial Differential Equations 25 (2000), no. 11-12, 2119--2169) prove that the solution exists in time $[0,T_\varepsilon)$ with $T_\varepsilon\ge Ce^{C\varepsilon^{-\mu}}$ ($\mu=1$ if $d\ge3$, $\mu=2/3$ if $d=2$). In the present paper, we will focus on the general quasilinear Klein-Gordon equation without the null condition and further show that the existence time of the solution can be improved to $T_\varepsilon=+\infty$ if $d\geq3$ and $T_\varepsilon\ge e^{C\varepsilon^{-2}}$ if $d=2$. In addition, for $d=2$ and any fixed number $\alpha>0$, if the weighted $L^2$ norm of the initial data with the weight $(1+|x|)^\alpha$ is small, then the solution exists globally and scatters to a free solution. The arguments are based on the introduction of a good unknown, the Strichartz estimate, the weighted $L^2$-norm estimate and the resonance analysis.

Poisson surface reconstruction (PSR) remains a popular technique for reconstructing watertight surfaces from 3D point samples thanks to its efficiency, simplicity, and robustness. Yet, the existing PSR method and subsequent variants work only for oriented points. This paper intends to validate that an improved PSR, called iPSR, can completely eliminate the requirement of point normals and proceed in an iterative manner. In each iteration, iPSR takes as input point samples with normals directly computed from the surface obtained in the preceding iteration, and then generates a new surface with better quality. Extensive quantitative evaluation confirms that the new iPSR algorithm converges in 5-30 iterations even with randomly initialized normals. If initialized with a simple visibility based heuristic, iPSR can further reduce the number of iterations. We conduct comprehensive comparisons with PSR and other powerful implicit-function based methods. Finally, we confirm iPSR's effectiveness and scalability on the AIM@SHAPE dataset and challenging (indoor and outdoor) scenes. Code and data for this paper are at https://github.com/houfei0801/ipsr.

Aiming at developing intuitive and easy-to-use portrait editing tools, we propose a novel vectorization method that can automatically convert raster images into a 3-tier hierarchical representation. The base layer consists of a set of sparse diffusion curves (DC) which characterize salient geometric features and low-frequency colors and provide means for semantic color transfer and facial expression editing. The middle level encodes specular highlights and shadows to large and editable Poisson regions (PR) and allows the user to directly adjust illumination via tuning the strength and/or changing shape of PR. The top level contains two types of pixel-sized PRs for high-frequency residuals and fine details such as pimples and pigmentation. We also train a deep generative model that can produce high-frequency residuals automatically. Thanks to the meaningful organization of vector primitives, editing portraits becomes easy and intuitive. In particular, our method supports color transfer, facial expression editing, highlight and shadow editing and automatic retouching. Thanks to the linearity of the Laplace operator, we introduce alpha blending, linear dodge and linear burn to vector editing and show that they are effective in editing highlights and shadows. To quantitatively evaluate the results, we extend the commonly used FLIP metric (which measures differences between two images) by considering illumination. The new metric, called illumination-sensitive FLIP or IS-FLIP, can effectively capture the salient changes in color transfer results, and is more consistent with human perception than FLIP and other quality measures on portrait images. We evaluate our method on the FFHQR dataset and show that our method is effective for common portrait editing tasks, such as retouching, light editing, color transfer and expression editing. We will make the code and trained models publicly available.

We develop a new method for portrait image editing, which supports fine-grained editing of geometries, colors, lights and shadows using a single neural network model. We adopt a novel asymmetric conditional GAN architecture: the generators take the transformed conditional inputs, such as edge maps, color palette, sliders and masks, that can be directly edited by the user; the discriminators take the conditional inputs in the way that can guide controllable image generation more effectively. Taking color editing as an example, we feed color palettes (which can be edited easily) into the generator, and color maps (which contain positional information of colors) into the discriminator. We also design a region-weighted discriminator so that higher weights are assigned to more important regions, like eyes and skin. Using a color palette, the user can directly specify the desired colors of hair, skin, eyes, lip and background. Color sliders allow the user to blend colors in an intuitive manner. The user can also edit lights and shadows by modifying the corresponding masks. We demonstrate the effectiveness of our method by evaluating it on the CelebAMask-HQ dataset with a wide range of tasks, including geometry/color/shadow/light editing, hand-drawn sketch to image translation, and color transfer. We also present ablation studies to justify our design.

For the 3D compressible isentropic Euler equations with an initial perturbation of size $\ve$ of a rest state, if the initial vorticity is of size $\dl$ with $0<\dl\le \ve$ and $\ve$ is small, we establish that the lifespan of the smooth solutions is $T_{\dl}=O(\min\{e^\frac{1}{\ve},\frac{1}{\delta}\})$ for the polytropic gases, and $T_{\dl}=O(\frac{1}{\delta})$ for the Chaplygin gases. For example, when $\dl=e^{-\f{1}{\ve^2}}$ is chosen, then $T_{\dl}=O(e^{\f{1}{\ve}})$ for the polytropic gases and $T_{\dl}=O(e^{\f{1}{\ve^2}})$ for the Chaplygin gases although the perturbations of the initial density and the divergence of the initial velocity are only of order $O(\ve)$. Our result illustrates that the time of existence of smooth solutions depends crucially on the size of the vorticity of the initial data, as long as the initial data is sufficiently close to a constant. The main ingredients in the paper are: introducing some suitably weighted energies, deriving the pointwise space-time decay estimates of solutions, looking for the good unknown instead of the velocity, and establishing the required weighted estimates on the vorticty.

For the 2D compressible isentropic Euler equations of polytropic gases with an initial perturbation of size $\ve$ of a rest state, it has been known that if the initial data are rotationnally invariant or irrotational, then the lifespan $T_{\ve}$ of the classical solutions is of order $O(\f{1}{\ve^2})$; if the initial vorticity is of size $\ve^{1+\al}$ ($0\le\al\le 1$), then $T_{\ve}$ is of $O(\f{1}{\ve^{1+\al}})$. In the present paper, for the 2D compressible isentropic Euler equations of Chaplygin gases, if the initial data are a perturbation of size $\ve$, and the initial vorticity is of any size $\dl$ with $0<\dl\le \ve$, we will establish the lifespan $T_{\dl}=O(\f{1}{\dl})$. For examples, if $\dl=e^{-\f{1}{\ve^2}}$ or $\dl=e^{-e^{\f{1}{\ve^2}}}$ are chosen, then $T_{\dl}=O(e^{\f{1}{\ve^2}})$ or $T_{\dl}=O(e^{e^{\f{1}{\ve^2}}})$ although the perturbations of the initial density and the divergence of the initial velocity are only of order $O(\ve)$. Our main ingredients are: finding the null condition structures in 2D compressible Euler equations of Chaplygin gases and looking for the good unknown; establishing a new class of weighted space-time $L^\infty$-$L^\infty$ estimates for the solution itself and its gradients of 2D linear wave equations; introducing some suitably weighted energies and taking the $L^p$ $(1