NeRF-Art: Text-Driven Neural Radiance Fields Stylization.

The success of neural fields for 3D vision tasks is now indisputable. Following this trend, several methods aiming for visual localization (e.g., SLAM) have been proposed to estimate distance or density fields using neural fields. However, it is difficult to achieve high localization performance by only density fields-based methods such as Neural Radiance Field (NeRF) since they do not provide density gradient in most empty regions. On the other hand, distance field-based methods such as Neural Implicit Surface (NeuS) have limitations in objects’ surface shapes. This paper proposes Neural Distance-Density Field (NeDDF), a novel 3D representation that reciprocally constrains the distance and density fields. We extend distance field formulation to shapes with no explicit boundary surface, such as fur or smoke, which enable explicit conversion from distance field to density field. Consistent distance and density fields realized by explicit conversion enable both robustness to initial values and high-quality registration. Furthermore, the consistency between fields allows fast convergence from sparse point clouds. Experiments show that NeDDF can achieve high localization performance while providing comparable results to NeRF on novel view synthesis. The code is available at https://github.com/ueda0319/neddf.

Novel view synthesis of remote sensing scenes is of great significance for scene visualization, human-computer interaction, and various downstream applications. Despite the recent advances in computer graphics and photogrammetry technology, generating novel views is still challenging particularly for remote sensing images due to its high complexity, view sparsity and limited view-perspective variations. In this paper, we propose a novel remote sensing view synthesis method by leveraging the recent advances in implicit neural representations. Considering the overhead and far depth imaging of remote sensing images, we represent the 3D space by combining implicit multiplane images (MPI) representation and deep neural networks. The 3D scene is reconstructed under a self-supervised optimization paradigm through a differentiable multiplane renderer with multi-view input constraints. Images from any novel views thus can be freely rendered on the basis of the reconstructed model. As a by-product, the depth maps corresponding to the given viewpoint can be generated along with the rendering output. We refer to our method as Implicit Multiplane Images (ImMPI). To further improve the view synthesis under sparse-view inputs, we explore the learning-based initialization of remote sensing 3D scenes and proposed a neural network based Prior extractor to accelerate the optimization process. In addition, we propose a new dataset for remote sensing novel view synthesis with multi-view real-world google earth images. Extensive experiments demonstrate the superiority of the ImMPI over previous state-of-the-art methods in terms of reconstruction accuracy, visual fidelity, and time efficiency. Ablation experiments also suggest the effectiveness of our methodology design. Our dataset and code can be found at https://github.com/wyc-Chang/ImMPI

3D Gaussian Splatting (3DGS) has recently made notable progress in photo-realistic view synthesis and 3D reconstruction from multi-view images. It initializes anisotropic Gaussian representations of complex scenes using sparse point clouds and optimizes these Gaussians using multiple perspective images. To apply 3DGS to panoramic images, we propose a panoramic view synthesis method. Specifically, each panoramic image is divided into multiple perspective images to obtain sparse point clouds and camera parameters through SfM. However, this division introduces stitching artifacts. To tackle this issue, we utilize a pre-trained depth estimation model to estimate depth as geometric guidance for depth supervision, where the inverse depth of each frame's SfM points is used to resolve scale and offset issues. We verify the proposed method on a custom indoor panoramic dataset. Compared to the original 3DGS, our method achieves relatively robust geometric structures and effectively reduces some artifacts. In addition, our method has the same effect after 7K iterations as 3DGS does after 30K iterations, and uses fewer Gaussians to represent the scene.

Novel view synthesis from a single image has recently attracted a lot of attention, and it has been primarily advanced by 3D deep learning and rendering techniques. However, most work is still limited by synthesizing new views within relatively small camera motions. In this paper, we propose a novel approach to synthesize a consistent long-term video given a single scene image and a trajectory of large camera motions. Our approach utilizes an autoregressive Transformer to perform sequential modeling of multiple frames, which reasons the relations between multiple frames and the corresponding cameras to predict the next frame. To facilitate learning and ensure consistency among generated frames, we introduce a locality constraint based on the input cameras to guide self-attention among a large number of patches across space and time. Our method outperforms state-of-the-art view synthesis approaches by a large margin, especially when synthesizing long-term future in indoor 3D scenes. Project page at https://xrenaa.github.io/look-outside-room/.

Recent advancements in 4D scene reconstruction using neural radiance fields (NeRF) have demonstrated the ability to represent dynamic scenes from multi-view videos. However, they fail to reconstruct the dynamic scenes and struggle to fit even the training views in unsynchronized settings. It happens because they employ a single latent embedding for a frame while the multi-view images at the same frame were actually captured at different moments. To address this limitation, we introduce time offsets for individual unsynchronized videos and jointly optimize the offsets with NeRF. By design, our method is applicable for various baselines and improves them with large margins. Furthermore, finding the offsets always works as synchronizing the videos without manual effort. Experiments are conducted on the common Plenoptic Video Dataset and a newly built Unsynchronized Dynamic Blender Dataset to verify the performance of our method. Project page: https://seoha-kim.github.io/sync-nerf

Neural implicit 3D representations have emerged as a powerful paradigm for reconstructing surfaces from multi-view images and synthesizing novel views. Unfortunately, existing methods such as DVR or IDR require accurate per-pixel object masks as supervision. At the same time, neural radiance fields have revolutionized novel view synthesis. However, NeRF’s estimated volume density does not admit accurate surface reconstruction. Our key insight is that implicit surface models and radiance fields can be formulated in a unified way, enabling both surface and volume rendering using the same model. This unified perspective enables novel, more efficient sampling procedures and the ability to reconstruct accurate surfaces without input masks. We compare our method on the DTU, BlendedMVS, and a synthetic indoor dataset. Our experiments demonstrate that we outperform NeRF in terms of reconstruction quality while performing on par with IDR without requiring masks.

To plan and execute robotics tasks, 3D scene perception and understanding are critical for robots to interact with complex environments effectively. Traditional systems often rely on closed vocabularies and are constrained by pre-defined object categories, which limit their flexibility. In this paper, we propose 3DS-Plan, an open-vocabulary 3D scene perception and generation model designed to facilitate long-horizon task planning. It leverages the pre-trained foundation models to support robotic scene understanding and further provides environmental details to infer actionable steps in various scenarios. We validate the effectiveness of our frame-work through comprehensive experiments. It demonstrates a 31.41 % improvement in computational efficiency for class operations and at least a 14.52 % increase in success rates for complex tasks without requiring extensive retraining or extra annotation. The corresponding project page is available at https://techpage.github.i0/open3dsp/.

While surface-based view synthesis algorithms are appealing due to their low computational requirements, they often struggle to reproduce thin structures. In contrast, more expensive methods that model the scene's geometry as a volumetric density field (e.g. NeRF) excel at reconstructing fine geometric detail. However, density fields often represent geometry in a "fuzzy" manner, which hinders exact localization of the surface. In this work, we modify density fields to encourage them to converge towards surfaces, without compromising their ability to reconstruct thin structures. First, we employ a discrete opacity grid representation instead of a continuous density field, which allows opacity values to discontinuously transition from zero to one at the surface. Second, we anti-alias by casting multiple rays per pixel, which allows occlusion boundaries and subpixel structures to be modelled without using semi-transparent voxels. Third, we minimize the binary entropy of the opacity values, which facilitates the extraction of surface geometry by encouraging opacity values to binarize towards the end of training. Lastly, we develop a fusion-based meshing strategy followed by mesh simplification and appearance model fitting. The compact meshes produced by our model can be rendered in real-time on mobile devices and achieve significantly higher view synthesis quality compared to existing mesh-based approaches. Our interactive webdemo is available at https://binary-opacity-grid.github.io.

Rethinking scene representation: A saliency-driven hierarchical multi-scale resampling for RGB-D scene point cloud in robotic applications

Radiance field based methods have recently revolutionized novel view synthesis of scenes captured with multi-view photos. A significant recent advance is 3D Gaussian Splatting (3DGS), which utilizes a set of 3D Gaussians to represent a radiance field, yielding high-fidelity results in real-time rendering. However, we have observed that 3DGS struggles to capture the necessary details in sparsely observed regions, where there is not enough gradient for effective split and clone operations. In this paper, we present a novel solution to address this limitation. Our key idea is to leverage segmentation information to identify poorly optimized regions within the 3D Gaussian representation. By applying split or clone operations on the corresponding 3D Gaussians in these regions, we aim to refine the spatial distribution of Gaussians and enhance the overall quality of high-fidelity 3D scene reconstruction. To further optimize the reconstruction process, we introduce two spatial regularization terms: repulsion loss and smoothness loss. These terms effectively minimize overlap and redundancy among Gaussians, reducing outliers in the synthesized geometry. By incorporating these regularization techniques, our approach achieves state-of-the-art performance in real-time novel view synthesis and significantly improves visibility in less observed regions, leading to a more compact and accurate 3D scene representation.

In this work, we address the lack of 3D understanding of generative neural networks by introducing a persistent 3D feature embedding for view synthesis. To this end, we propose DeepVoxels, a learned representation that encodes the view-dependent appearance of a 3D scene without having to explicitly model its geometry. At its core, our approach is based on a Cartesian 3D grid of persistent embedded features that learn to make use of the underlying 3D scene structure. Our approach combines insights from 3D geometric computer vision with recent advances in learning image-to-image mappings based on adversarial loss functions. DeepVoxels is supervised, without requiring a 3D reconstruction of the scene, using a 2D re-rendering loss and enforces perspective and multi-view geometry in a principled manner. We apply our persistent 3D scene representation to the problem of novel view synthesis demonstrating high-quality results for a variety of challenging scenes.

Neural radiance fields (NeRF) have achieved great success in novel view synthesis and 3D representation for static scenarios. Existing dynamic NeRFs usually exploit a locally dense grid to fit the deformation fields; however, they fail to capture the global dynamics and concomitantly yield models of heavy parameters. We observe that the 4D space is inherently sparse. First, the deformation fields are sparse in spatial but dense in temporal due to the continuity of motion. Second, the radiance fields are only valid on the surface of the underlying scene, usually occupying a small fraction of the whole space. We thus represent the 4D scene using a learnable sparse latent space, a.k.a. SLS4D. Specifically, SLS4D first uses dense learnable time slot features to depict the temporal space, from which the deformation fields are fitted with linear multi-layer perceptions (MLP) to predict the displacement of a 3D position at any time. It then learns the spatial features of a 3D position using another sparse latent space. This is achieved by learning the adaptive weights of each latent feature with the attention mechanism. Extensive experiments demonstrate the effectiveness of our SLS4D: It achieves the best 4D novel view synthesis using only about 6% parameters of the most recent work.

In this paper, we present WonderHuman to reconstruct dynamic human avatars from a monocular video for high-fidelity novel view synthesis. Previous dynamic human avatar reconstruction methods typically require the input video to have full coverage of the observed human body. However, in daily practice, one typically has access to limited viewpoints, such as monocular front-view videos, making it a cumbersome task for previous methods to reconstruct the unseen parts of the human avatar. To tackle the issue, we present WonderHuman, which leverages 2D generative diffusion model priors to achieve high-quality, photorealistic reconstructions of dynamic human avatars from monocular videos, including accurate rendering of unseen body parts. Our approach introduces a Dual-Space Optimization technique, applying Score Distillation Sampling (SDS) in both canonical and observation spaces to ensure visual consistency and enhance realism in dynamic human reconstruction. Additionally, we present a View Selection strategy and Pose Feature Injection to enforce the consistency between SDS predictions and observed data, ensuring pose-dependent effects and higher fidelity in the reconstructed avatar. In the experiments, our method achieves SOTA performance in producing photorealistic renderings from the given monocular video, particularly for those challenging unseen parts.

Recent advances in novel view synthesis have demonstrated impressive results in fast photorealistic scene rendering through differentiable point rendering, either via Gaussian Splatting (3DGS) [Kerbl et al. 2023 ] or neural point rendering [Aliev et al. 2020 ]. Unfortunately, these directions require either a large number of small Gaussians or expensive per-pixel post-processing for reconstructing fine details, which negatively impacts rendering performance. To meet the high performance demands of virtual reality (VR) systems, primitive or pixel counts therefore must be kept low, affecting visual quality. In this paper, we propose a novel hybrid approach based on foveated rendering as a promising solution that combines the strengths of both point rendering directions regarding performance sweet spots. Analyzing the compatibility with the human visual system, we find that using a low-detailed, few primitive smooth Gaussian representation for the periphery is cheap to compute and meets the perceptual demands of peripheral vision. For the fovea only, we use neural points with a convolutional neural network for the small pixel footprint, which provides sharp, detailed output within the rendering budget. This combination also allows for synergistic method accelerations with point occlusion culling and reducing the demands on the neural network. Our evaluation confirms that our approach increases sharpness and details compared to a standard VR-ready 3DGS configuration, and participants of a user study overwhelmingly preferred our method. Our system meets the necessary performance requirements for real-time VR interactions, ultimately enhancing the user’s immersive experience. The project page can be found at: https://lfranke.github.io/vr_splatting

Differentiable rendering techniques have recently shown promising results for free-viewpoint video synthesis of characters. However, such methods, either Gaussian Splatting or neural implicit rendering, typically necessitate per-subject optimization which does not meet the requirement of real-time rendering in an interactive application. We propose a generalizable Gaussian Splatting approach for high-resolution image rendering under a sparse-view camera setting. To this end, we introduce Gaussian parameter maps defined on the source views and directly regress Gaussian properties for instant novel view synthesis without any fine-tuning or optimization. We train our Gaussian parameter regression module on human-only data or human-scene data, jointly with a depth estimation module to lift 2D parameter maps to 3D space. The proposed framework is fully differentiable with both depth and rendering supervision or with only rendering supervision. We further introduce a regularization term and an epipolar attention mechanism to preserve geometry consistency between two source views, especially when neglecting depth supervision. Experiments on several datasets demonstrate that our method outperforms state-of-the-art methods while achieving an exceeding rendering speed. Our project page is available at https://yaourtb.github.io/GPS-Gaussian+.