cs.GR

arXiv:2501.01424v1 Announce Type: new Abstract: We introduce a method for composing object-level visual prompts within a text-to-image diffusion model. Our approach addresses the task of generating semantically coherent compositions across diverse scenes and styles, similar to the versatility and expressiveness offered by text prompts. A key challenge in this task is to preserve the identity of the objects depicted in the input visual prompts, while also generating diverse compositions across different images. To address this challenge, we introduce a new KV-mixed cross-attention mechanism, in which keys and values are learned from distinct visual representations. The keys are derived from an encoder with a small bottleneck for layout control, whereas the values come from a larger bottleneck encoder that captures fine-grained appearance details. By mixing keys and values from these complementary sources, our model preserves the identity of the visual prompts while supporting flexible variations in object arrangement, pose, and composition. During inference, we further propose object-level compositional guidance to improve the method's identity preservation and layout correctness. Results show that our technique produces diverse scene compositions that preserve the unique characteristics of each visual prompt, expanding the creative potential of text-to-image generation.

arXiv:2501.01393v1 Announce Type: new Abstract: Garment animation is ubiquitous in various applications, such as virtual reality, gaming, and film producing. Recently, learning-based approaches obtain compelling performance in animating diverse garments under versatile scenarios. Nevertheless, to mimic the deformations of the observed garments, data-driven methods require large scale of garment data, which are both resource-wise expensive and time-consuming. In addition, forcing models to match the dynamics of observed garment animation may hinder the potentials to generalize to unseen cases. In this paper, instead of using garment-wise supervised-learning we adopt a disentangled scheme to learn how to animate observed garments: 1). learning constitutive behaviors from the observed cloth; 2). dynamically animate various garments constrained by the learned constitutive laws. Specifically, we propose Energy Unit network (EUNet) to model the constitutive relations in the format of energy. Without the priors from analytical physics models and differentiable simulation engines, EUNet is able to directly capture the constitutive behaviors from the observed piece of cloth and uniformly describes the change of energy caused by deformations, such as stretching and bending. We further apply the pre-trained EUNet to animate various garments based on energy optimizations. The disentangled scheme alleviates the need of garment data and enables us to utilize the dynamics of a piece of cloth for animating garments. Experiments show that while EUNet effectively delivers the energy gradients due to the deformations, models constrained by EUNet achieve more stable and physically plausible performance comparing with those trained in garment-wise supervised manner. Code is available at https://github.com/ftbabi/EUNet_NeurIPS2024.git .

arXiv:2409.03164v2 Announce Type: replace Abstract: The high performance of tree ensemble classifiers benefits from a large set of rules, which, in turn, makes the models hard to understand. To improve interpretability, existing methods extract a subset of rules for approximation using model reduction techniques. However, by focusing on the reduced rule set, these methods often lose fidelity and ignore anomalous rules that, despite their infrequency, play crucial roles in real-world applications. This paper introduces a scalable visual analysis method to explain tree ensemble classifiers that contain tens of thousands of rules. The key idea is to address the issue of losing fidelity by adaptively organizing the rules as a hierarchy rather than reducing them. To ensure the inclusion of anomalous rules, we develop an anomaly-biased model reduction method to prioritize these rules at each hierarchical level. Synergized with this hierarchical organization of rules, we develop a matrix-based hierarchical visualization to support exploration at different levels of detail. Our quantitative experiments and case studies demonstrate how our method fosters a deeper understanding of both common and anomalous rules, thereby enhancing interpretability without sacrificing comprehensiveness.

arXiv:2401.12977v2 Announce Type: replace Abstract: Inverse rendering seeks to recover 3D geometry, surface material, and lighting from captured images, enabling advanced applications such as novel-view synthesis, relighting, and virtual object insertion. However, most existing techniques rely on high dynamic range (HDR) images as input, limiting accessibility for general users. In response, we introduce IRIS, an inverse rendering framework that recovers the physically based material, spatially-varying HDR lighting, and camera response functions from multi-view, low-dynamic-range (LDR) images. By eliminating the dependence on HDR input, we make inverse rendering technology more accessible. We evaluate our approach on real-world and synthetic scenes and compare it with state-of-the-art methods. Our results show that IRIS effectively recovers HDR lighting, accurate material, and plausible camera response functions, supporting photorealistic relighting and object insertion.

arXiv:2501.01407v1 Announce Type: new Abstract: Personalizing text-to-image models to generate images of specific subjects across diverse scenes and styles is a rapidly advancing field. Current approaches often face challenges in maintaining a balance between identity preservation and alignment with the input text prompt. Some methods rely on a single textual token to represent a subject, which limits expressiveness, while others employ richer representations but disrupt the model's prior, diminishing prompt alignment. In this work, we introduce Nested Attention, a novel mechanism that injects a rich and expressive image representation into the model's existing cross-attention layers. Our key idea is to generate query-dependent subject values, derived from nested attention layers that learn to select relevant subject features for each region in the generated image. We integrate these nested layers into an encoder-based personalization method, and show that they enable high identity preservation while adhering to input text prompts. Our approach is general and can be trained on various domains. Additionally, its prior preservation allows us to combine multiple personalized subjects from different domains in a single image.

arXiv:2501.00601v1 Announce Type: new Abstract: Synthesizing photo-realistic visual observations from an ego vehicle's driving trajectory is a critical step towards scalable training of self-driving models. Reconstruction-based methods create 3D scenes from driving logs and synthesize geometry-consistent driving videos through neural rendering, but their dependence on costly object annotations limits their ability to generalize to in-the-wild driving scenarios. On the other hand, generative models can synthesize action-conditioned driving videos in a more generalizable way but often struggle with maintaining 3D visual consistency. In this paper, we present DreamDrive, a 4D spatial-temporal scene generation approach that combines the merits of generation and reconstruction, to synthesize generalizable 4D driving scenes and dynamic driving videos with 3D consistency. Specifically, we leverage the generative power of video diffusion models to synthesize a sequence of visual references and further elevate them to 4D with a novel hybrid Gaussian representation. Given a driving trajectory, we then render 3D-consistent driving videos via Gaussian splatting. The use of generative priors allows our method to produce high-quality 4D scenes from in-the-wild driving data, while neural rendering ensures 3D-consistent video generation from the 4D scenes. Extensive experiments on nuScenes and street view images demonstrate that DreamDrive can generate controllable and generalizable 4D driving scenes, synthesize novel views of driving videos with high fidelity and 3D consistency, decompose static and dynamic elements in a self-supervised manner, and enhance perception and planning tasks for autonomous driving.

arXiv:2501.01382v1 Announce Type: new Abstract: The head-mounted display is a vital component of augmented reality, incorporating optics with complex display and see-through optical behavior. Computationally modeling these optical behaviors requires meeting three key criteria: accuracy, efficiency, and accessibility. In recent years, various approaches have been proposed to model display and see-through optics, which can broadly be classified into black-box and white-box models. However, both categories face significant limitations that hinder their adoption in commercial applications. To overcome these challenges, we leveraged prior knowledge of ray bundle properties outside the optical hardware and proposed a novel bundle-fit-based model. In this approach, the ray paths within the optics are treated as a black box, while a lightweight optimization problem is solved to fit the ray bundle outside the optics. This method effectively addresses the accuracy issues of black-box models and the accessibility challenges of white-box models. Although our model involves runtime optimization, this is typically not a concern, as it can use the solution from a previous query to initialize the optimization for the current query. We evaluated the performance of our proposed method through both simulations and experiments on real hardware, demonstrating its effectiveness.

arXiv:2501.00625v1 Announce Type: new Abstract: Recently released open-source pre-trained foundational image segmentation and object detection models (SAM2+GroundingDINO) allow for geometrically consistent segmentation of objects of interest in multi-view 2D images. Users can use text-based or click-based prompts to segment objects of interest without requiring labeled training datasets. Gaussian Splatting allows for the learning of the 3D representation of a scene's geometry and radiance based on 2D images. Combining Google Earth Studio, SAM2+GroundingDINO, 2D Gaussian Splatting, and our improvements in mask refinement based on morphological operations and contour simplification, we created a pipeline to extract the 3D mesh of any building based on its name, address, or geographic coordinates.

arXiv:2501.01362v1 Announce Type: new Abstract: Complex geometric tasks such as geometric modeling, physical simulation, and texture parametrization often involve the embedding of many complex sub-domains with potentially different dimensions. These tasks often require evolving the geometry and topology of the discretizations of these sub-domains, and guaranteeing a \emph{consistent} overall embedding for the multiplicity of sub-domains is required to define boundary conditions. We propose a data structure and algorithmic framework for hierarchically encoding a collection of meshes, enabling topological and geometric changes to be automatically propagated with coherent correspondences between them. We demonstrate the effectiveness of our approach in surface mesh decimation while preserving UV seams, periodic 2D/3D meshing, and extending the TetWild algorithm to ensure topology preservation of the embedded structures.

arXiv:2412.10748v2 Announce Type: replace Abstract: Simulating fuel sloshing within aircraft tanks during flight is crucial for aircraft safety research. Traditional methods based on Navier-Stokes equations are computationally expensive. In this paper, we treat fluid motion as point cloud transformation and propose the first neural network method specifically designed for simulating fuel sloshing in aircraft. This model is also the deep learning model that is the first to be capable of stably modeling fluid particle dynamics in such complex scenarios. Our triangle feature fusion design achieves an optimal balance among fluid dynamics modeling, momentum conservation constraints, and global stability control. Additionally, we constructed the Fueltank dataset, the first dataset for aircraft fuel surface sloshing. It comprises 320,000 frames across four typical tank types and covers a wide range of flight maneuvers, including multi-directional rotations. We conducted comprehensive experiments on both our dataset and the take-off scenario of the aircraft. Compared to existing neural network-based fluid simulation algorithms, we significantly enhanced accuracy while maintaining high computational speed. Compared to traditional SPH methods, our speed improved approximately 10 times. Furthermore, compared to traditional fluid simulation software such as Flow3D, our computation speed increased by more than 300 times.

arXiv:2412.18380v1 Announce Type: new Abstract: This study presents RSGaussian, an innovative novel view synthesis (NVS) method for aerial remote sensing scenes that incorporate LiDAR point cloud as constraints into the 3D Gaussian Splatting method, which ensures that Gaussians grow and split along geometric benchmarks, addressing the overgrowth and floaters issues occurs. Additionally, the approach introduces coordinate transformations with distortion parameters for camera models to achieve pixel-level alignment between LiDAR point clouds and 2D images, facilitating heterogeneous data fusion and achieving the high-precision geo-alignment required in aerial remote sensing. Depth and plane consistency losses are incorporated into the loss function to guide Gaussians towards real depth and plane representations, significantly improving depth estimation accuracy. Experimental results indicate that our approach has achieved novel view synthesis that balances photo-realistic visual quality and high-precision geometric estimation under aerial remote sensing datasets. Finally, we have also established and open-sourced a dense LiDAR point cloud dataset along with its corresponding aerial multi-view images, AIR-LONGYAN.

arXiv:2407.07755v2 Announce Type: replace Abstract: Neural surfaces (e.g., neural map encoding, deep implicits 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. In the absence of 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 [Morreale et al. 2021]. 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 that this work can become a key ingredient in enabling neural geometry processing. Code will be released upon acceptance, accessible from the project webpage.

arXiv:2412.17957v1 Announce Type: new Abstract: $\textit{ArchComplete}$ is a two-stage dense voxel-based 3D generative pipeline developed to tackle the high complexity in architectural geometries and topologies, assisting with ideation and geometric detailisation in the early design process. In stage 1, a $\textit{3D Voxel VQGAN}$ model is devised, whose composition is then modelled with an autoregressive transformer for generating coarse models. Subsequently, in stage 2, $\textit{Hierarchical Voxel Upsampling Networks}$ consisting of a set of 3D conditional denoising diffusion probabilistic models are defined to augment the coarse shapes with fine geometric details. The first stage is trained on a dataset of house models with fully modelled exteriors and interiors with a novel 2.5D perceptual loss to capture input complexities across multiple abstraction levels, while the second stage trains on randomly cropped local volumetric patches, requiring significantly less compute and memory. For inference, the pipeline first autoregressively generates house models at a resolution of $64^3$ and then progressively refines them to resolution of $256^3$ with voxel sizes as small as $18\text{cm}$. ArchComplete supports a range of interaction modes solving a variety of tasks, including interpolation, variation generation, unconditional synthesis, and two conditional synthesis tasks: shape completion and plan-drawing completion, as well as geometric detailisation. The results demonstrate notable improvements against state-of-the-art on established metrics.

arXiv:2412.18600v1 Announce Type: new Abstract: Human-scene interaction (HSI) generation is crucial for applications in embodied AI, virtual reality, and robotics. While existing methods can synthesize realistic human motions in 3D scenes and generate plausible human-object interactions, they heavily rely on datasets containing paired 3D scene and motion capture data, which are expensive and time-consuming to collect across diverse environments and interactions. We present ZeroHSI, a novel approach that enables zero-shot 4D human-scene interaction synthesis by integrating video generation and neural human rendering. Our key insight is to leverage the rich motion priors learned by state-of-the-art video generation models, which have been trained on vast amounts of natural human movements and interactions, and use differentiable rendering to reconstruct human-scene interactions. ZeroHSI can synthesize realistic human motions in both static scenes and environments with dynamic objects, without requiring any ground-truth motion data. We evaluate ZeroHSI on a curated dataset of different types of various indoor and outdoor scenes with different interaction prompts, demonstrating its ability to generate diverse and contextually appropriate human-scene interactions.

arXiv:2412.18086v1 Announce Type: new Abstract: Motion planning is a crucial component in autonomous driving. State-of-the-art motion planners are trained on meticulously curated datasets, which are not only expensive to annotate but also insufficient in capturing rarely seen critical scenarios. Failing to account for such scenarios poses a significant risk to motion planners and may lead to incidents during testing. An intuitive solution is to manually compose such scenarios by programming and executing a simulator (e.g., CARLA). However, this approach incurs substantial human costs. Motivated by this, we propose an inexpensive method for generating diverse critical traffic scenarios to train more robust motion planners. First, we represent traffic scenarios as scripts, which are then used by the simulator to generate traffic scenarios. Next, we develop a method that accepts user-specified text descriptions, which a Large Language Model (LLM) translates into scripts using in-context learning. The output scripts are sent to the simulator that produces the corresponding traffic scenarios. As our method can generate abundant safety-critical traffic scenarios, we use them as synthetic training data for motion planners. To demonstrate the value of generated scenarios, we train existing motion planners on our synthetic data, real-world datasets, and a combination of both. Our experiments show that motion planners trained with our data significantly outperform those trained solely on real-world data, showing the usefulness of our synthetic data and the effectiveness of our data generation method. Our source code is available at https://ezharjan.github.io/AutoSceneGen.

arXiv:2412.16242v1 Announce Type: new Abstract: Transparency is commonly utilized in visualizations to overlay color-coded histograms or sets, thereby facilitating the visual comparison of categorical data. However, these charts often suffer from significant overlap between objects, resulting in substantial color interactions. Existing color blending models struggle in these scenarios, frequently leading to ambiguous color mappings and the introduction of false colors. To address these challenges, we propose an automated approach for generating optimal color encodings to enhance the perception of translucent charts. Our method harnesses color nameability to maximize the association between composite colors and their respective class labels. We introduce a color-name aware (CNA) optimization framework that generates maximally coherent color assignments and transparency settings while ensuring perceptual discriminability for all segments in the visualization. We demonstrate the effectiveness of our technique through crowdsourced experiments with composite histograms, showing how our technique can significantly outperform both standard and visualization-specific color blending models. Furthermore, we illustrate how our approach can be generalized to other visualizations, including parallel coordinates and Venn diagrams. We provide an open-source implementation of our technique as a web-based tool.

arXiv:2412.16211v1 Announce Type: new Abstract: The current state-of-the-art video generative models can produce commercial-grade videos with highly realistic details. However, they still struggle to coherently present multiple sequential events in the stories specified by the prompts, which is foreseeable an essential capability for future long video generation scenarios. For example, top T2V generative models still fail to generate a video of the short simple story 'how to put an elephant into a refrigerator.' While existing detail-oriented benchmarks primarily focus on fine-grained metrics like aesthetic quality and spatial-temporal consistency, they fall short of evaluating models' abilities to handle event-level story presentation. To address this gap, we introduce StoryEval, a story-oriented benchmark specifically designed to assess text-to-video (T2V) models' story-completion capabilities. StoryEval features 423 prompts spanning 7 classes, each representing short stories composed of 2-4 consecutive events. We employ advanced vision-language models, such as GPT-4V and LLaVA-OV-Chat-72B, to verify the completion of each event in the generated videos, applying a unanimous voting method to enhance reliability. Our methods ensure high alignment with human evaluations, and the evaluation of 11 models reveals its challenge, with none exceeding an average story-completion rate of 50%. StoryEval provides a new benchmark for advancing T2V models and highlights the challenges and opportunities in developing next-generation solutions for coherent story-driven video generation.

arXiv:2412.16213v1 Announce Type: new Abstract: The increasing deployment of AI models in critical applications has exposed them to significant risks from adversarial attacks. While adversarial vulnerabilities in 2D vision models have been extensively studied, the threat landscape for 3D generative models, such as Neural Radiance Fields (NeRF), remains underexplored. This work introduces \textit{AdvIRL}, a novel framework for crafting adversarial NeRF models using Instant Neural Graphics Primitives (Instant-NGP) and Reinforcement Learning. Unlike prior methods, \textit{AdvIRL} generates adversarial noise that remains robust under diverse 3D transformations, including rotations and scaling, enabling effective black-box attacks in real-world scenarios. Our approach is validated across a wide range of scenes, from small objects (e.g., bananas) to large environments (e.g., lighthouses). Notably, targeted attacks achieved high-confidence misclassifications, such as labeling a banana as a slug and a truck as a cannon, demonstrating the practical risks posed by adversarial NeRFs. Beyond attacking, \textit{AdvIRL}-generated adversarial models can serve as adversarial training data to enhance the robustness of vision systems. The implementation of \textit{AdvIRL} is publicly available at \url{https://github.com/Tommy-Nguyen-cpu/AdvIRL/tree/MultiView-Clean}, ensuring reproducibility and facilitating future research.

arXiv:2412.16253v1 Announce Type: new Abstract: Generating high-quality 3D digital assets often requires expert knowledge of complex design tools. We introduce Specialized Generative Primitives, a generative framework that allows non-expert users to author high-quality 3D scenes in a seamless, lightweight, and controllable manner. Each primitive is an efficient generative model that captures the distribution of a single exemplar from the real world. With our framework, users capture a video of an environment, which we turn into a high-quality and explicit appearance model thanks to 3D Gaussian Splatting. Users then select regions of interest guided by semantically-aware features. To create a generative primitive, we adapt Generative Cellular Automata to single-exemplar training and controllable generation. We decouple the generative task from the appearance model by operating on sparse voxels and we recover a high-quality output with a subsequent sparse patch consistency step. Each primitive can be trained within 10 minutes and used to author new scenes interactively in a fully compositional manner. We showcase interactive sessions where various primitives are extracted from real-world scenes and controlled to create 3D assets and scenes in a few minutes. We also demonstrate additional capabilities of our primitives: handling various 3D representations to control generation, transferring appearances, and editing geometries.

arXiv:2412.16461v1 Announce Type: new Abstract: We propose a parameter optimization method for achieving static equilibrium of discrete elastic rods. Our method simultaneously optimizes material stiffness and rest shape parameters under box constraints to exactly enforce zero net force while avoiding stability issues and violations of physical laws. For efficiency, we split our constrained optimization problem into primal and dual subproblems via the augmented Lagrangian method, while handling the dual subproblem via simple vector updates. To efficiently solve the box-constrained primal subproblem, we propose a new active-set Cholesky preconditioner. Our method surpasses prior work in generality, robustness, and speed.