Seamless integration of both aerial and street view images remains a significant challenge in neural scene reconstruction and rendering. Existing methods predominantly focus on single domain, limiting their applications in immersive environments, which demand extensive free view exploration with large view changes both horizontally and vertically. We introduce Horizon-Gs, a novel approach built upon Gaussian Splatting techniques, tackles the unified reconstruction and rendering for aerial and street views. Our method addresses the key challenges of combining these perspectives with a new training strategy, overcoming viewpoint discrepancies to generate high-fidelity scenes. We also curate a high-quality aerial-to-ground views dataset encompassing both synthetic and real-world scene to advance further research. Experiments across diverse urban scene datasets confirm the effectiveness of our method.

The scaling law has been validated in various domains, such as natural language processing (NLP) and massive computer vision tasks; however, its application to motion generation remains largely unexplored. In this paper, we introduce a scalable motion generation framework that includes the motion tokenizer Motion FSQ-VAE and a text-prefix autoregressive transformer. Through comprehensive experiments, we observe the scaling behavior of this system. For the first time, we confirm the existence of scaling laws within the context of motion generation. Specifically, our results demonstrate that the normalized test loss of our prefix autoregressive models adheres to a logarithmic law in relation to compute budgets. Furthermore, we also confirm the power law between Non-Vocabulary Parameters, Vocabulary Parameters, and Data Tokens with respect to compute budgets respectively. Leveraging the scaling law, we predict the optimal transformer size, vocabulary size, and data requirements for a compute budget of 1e18. The test loss of the system, when trained with the optimal model size, vocabulary size, and required data, aligns precisely with the predicted test loss, thereby validating the scaling law. Project page: https://shunlinlu.github.io/ScaMo/

Recent advances in generative models have enabled high-quality 3D character reconstruction from multi-modal. However, animating these generated characters remains a challenging task, especially for complex elements like garments and hair, due to the lack of large-scale datasets and effective rigging methods. To address this gap, we curate AnimeRig, a large-scale dataset with detailed skeleton and skinning annotations. Building upon this, we propose DRiVE, a novel framework for generating and rigging 3D human characters with intricate structures. Unlike existing methods, DRiVE utilizes a 3D Gaussian representation, facilitating efficient animation and high-quality rendering. We further introduce GSDiff, a 3D Gaussian-based diffusion module that predicts joint positions as spatial distributions, overcoming the limitations of regression-based approaches. Extensive experiments demonstrate that DRiVE achieves precise rigging results, enabling realistic dynamics for clothing and hair, and surpassing previous methods in both quality and versatility. Code, dataset and visualization results are available at https://DRiVEAvatar.github.io/.

Generating diverse and natural human motion sequences based on textual descriptions constitutes a fundamental and challenging research area within the domains of computer vision, graphics, and robotics. Despite significant advancements in this field, current methodologies often face challenges regarding zero-shot generalization capabilities, largely attributable to the limited size of training datasets. Moreover, the lack of a comprehensive evaluation framework impedes the advancement of this task by failing to identify directions for improvement. In this work, we aim to push text-to-motion into a new era, that is, to achieve the generalization ability of zero-shot. To this end, firstly, we develop an efficient annotation pipeline and introduce MotionMillion-the largest human motion dataset to date, featuring over 2,000 hours and 2 million high-quality motion sequences. Additionally, we propose MotionMillion-Eval, the most comprehensive benchmark for evaluating zero-shot motion generation. Leveraging a scalable architecture, we scale our model to 7B parameters and validate its performance on MotionMillion-Eval. Our results demonstrate strong generalization to out-of-domain and complex compositional motions, marking a significant step toward zero-shot human motion generation. The code is available at https://github.com/VankouF/MotionMillion-Codes.

We present a unified and generalizable framework for synthesizing view-consistent and temporally coherent avatars from a single image, addressing the challenging task of single-image avatar generation. Existing diffusion-based methods often condition on sparse human templates (e.g., depth or normal maps), which leads to multi-view and temporal inconsistencies due to the mismatch between these signals and the true appearance of the subject. Our approach bridges this gap by combining the reconstruction power of regression-based 3D human reconstruction with the generative capabilities of a diffusion model. In a first step, an initial 3D reconstructed human through a generalized NeRF provides comprehensive conditioning, ensuring high-quality synthesis faithful to the reference appearance and structure. Subsequently, the derived geometry and appearance from the generalized NeRF serve as input to a video-based diffusion model. This strategic integration is pivotal for enforcing both multi-view and temporal consistency throughout the avatar's generation. Empirical results underscore the superior generalization ability of our proposed method, demonstrating its effectiveness across diverse in-domain and out-of-domain in-the-wild datasets.

Recent advancements in large-scale generative models have significantly improved the quality and diversity of 3D shape generation. However, most existing methods focus primarily on generating static 3D models, overlooking the potentially dynamic nature of certain shapes, such as humanoids, animals, and insects. To address this gap, we focus on rigging, a fundamental task in animation that establishes skeletal structures and skinning for 3D models. In this paper, we introduce OmniRig, the first large-scale rigging dataset, comprising 79,499 meshes with detailed skeleton and skinning information. Unlike traditional benchmarks that rely on predefined standard poses (e.g., A-pose, T-pose), our dataset embraces diverse shape categories, styles, and poses. Leveraging this rich dataset, we propose ARMO, a novel rigging framework that utilizes an autoregressive model to predict both joint positions and connectivity relationships in a unified manner. By treating the skeletal structure as a complete graph and discretizing it into tokens, we encode the joints using an auto-encoder to obtain a latent embedding and an autoregressive model to predict the tokens. A mesh-conditioned latent diffusion model is used to predict the latent embedding for conditional skeleton generation. Our method addresses the limitations of regression-based approaches, which often suffer from error accumulation and suboptimal connectivity estimation. Through extensive experiments on the OmniRig dataset, our approach achieves state-of-the-art performance in skeleton prediction, demonstrating improved generalization across diverse object categories. The code and dataset will be made public for academic use upon acceptance.

3D human digitization has long been a highly pursued yet challenging task. Existing methods aim to generate high-quality 3D digital humans from single or multiple views, but remain primarily constrained by current paradigms and the scarcity of 3D human assets. Specifically, recent approaches fall into several paradigms: optimization-based and feed-forward (both single-view regression and multi-view generation with reconstruction). However, they are limited by slow speed, low quality, cascade reasoning, and ambiguity in mapping low-dimensional planes to high-dimensional space due to occlusion and invisibility, respectively. Furthermore, existing 3D human assets remain small-scale, insufficient for large-scale training. To address these challenges, we propose a latent space generation paradigm for 3D human digitization, which involves compressing multi-view images into Gaussians via a UV-structured VAE, along with DiT-based conditional generation, we transform the ill-posed low-to-high-dimensional mapping problem into a learnable distribution shift, which also supports end-to-end inference. In addition, we employ the multi-view optimization approach combined with synthetic data to construct the HGS-1M dataset, which contains $1$ million 3D Gaussian assets to support the large-scale training. Experimental results demonstrate that our paradigm, powered by large-scale training, produces high-quality 3D human Gaussians with intricate textures, facial details, and loose clothing deformation.

This paper addresses the problem of decomposed 4D scene reconstruction from multi-view videos. Recent methods achieve this by lifting video segmentation results to a 4D representation through differentiable rendering techniques. Therefore, they heavily rely on the quality of video segmentation maps, which are often unstable, leading to unreliable reconstruction results. To overcome this challenge, our key idea is to represent the decomposed 4D scene with the Freetime FeatureGS and design a streaming feature learning strategy to accurately recover it from per-image segmentation maps, eliminating the need for video segmentation. Freetime FeatureGS models the dynamic scene as a set of Gaussian primitives with learnable features and linear motion ability, allowing them to move to neighboring regions over time. We apply a contrastive loss to Freetime FeatureGS, forcing primitive features to be close or far apart based on whether their projections belong to the same instance in the 2D segmentation map. As our Gaussian primitives can move across time, it naturally extends the feature learning to the temporal dimension, achieving 4D segmentation. Furthermore, we sample observations for training in a temporally ordered manner, enabling the streaming propagation of features over time and effectively avoiding local minima during the optimization process. Experimental results on several datasets show that the reconstruction quality of our method outperforms recent methods by a large margin.

In this paper, we focus on capturing closely interacted two-person motions from monocular videos, an important yet understudied topic. Unlike less-interacted motions, closely interacted motions contain frequently occurring inter-human occlusions, which pose significant challenges to existing capturing algorithms. To address this problem, our key observation is that close physical interactions between two subjects typically happen under very specific situations (e.g., handshake, hug, etc.), and such situational contexts contain strong prior semantics to help infer the poses of occluded joints. In this spirit, we introduce reaction priors, which are invertible neural networks that bi-directionally model the pose probability distributions of one person given the pose of the other. The learned reaction priors are then incorporated into a query-based pose estimator, which is a decoder-only Transformer with self-attentions on both intra-joint and inter-joint relationships. We demonstrate that our design achieves considerably higher performance than previous methods on multiple benchmarks. What's more, as existing datasets lack sufficient cases of close human-human interactions, we also build a new dataset called Dual-Human to better evaluate different methods. Dual-Human contains around 2k sequences of closely interacted two-person motions, each with synthetic multi-view renderings, contact annotations, and text descriptions. We believe that this new public dataset can significantly promote further research in this area. Our project page is at https://netease-gameai.github.io/Dual-Human/.

Generating multiview images from a single view facilitates the rapid generation of a 3D mesh conditioned on a single image. Recent methods [31] that introduce 3D global representation into diffusion models have shown the potential to generate consistent multiviews, but they have reduced generation speed and face challenges in maintaining generalizability and quality. To address this issue, we propose EpiDiff, a localized interactive multiview diffusion model. At the core of the proposed approach is to insert a lightweight epipolar attention block into the frozen diffusion model, leveraging epipolar constraints to enable cross-view interaction among feature maps of neighboring views. The newly initialized 3D modeling module preserves the original feature distribution of the diffusion model, exhibiting compatibility with a variety of base diffusion models. Experiments show that EpiDiff generates 16 multiview images in just 12 seconds, and it surpasses previous methods in quality evaluation metrics, including PSNR, SSIM and LPIPS. Additionally, EpiDiff can generate a more diverse distribution of views, improving the reconstruction quality from generated multiviews. Please see the project page at huanngzh.github.io/EpiDiff/.

This paper addresses the challenge of reconstructing an animatable human model from a multi-view video. Some recent works have proposed to decompose a non-rigidly deforming scene into a canonical neural radiance field and a set of deformation fields that map observation-space points to the canonical space, thereby enabling them to learn the dynamic scene from images. However, they represent the deformation field as translational vector field or SE(3) field, which makes the optimization highly under-constrained. Moreover, these representations cannot be explicitly controlled by input motions. Instead, we introduce blend weight fields to produce the deformation fields. Based on the skeleton-driven deformation, blend weight fields are used with 3D human skeletons to generate observation-to-canonical and canonical-to-observation correspondences. Since 3D human skeletons are more observable, they can regularize the learning of deformation fields. Moreover, the blend weight fields can be combined with input skeletal motions to generate new deformation fields to animate the human model. To improve the quality of human modeling, we further represent the human geometry as a signed distance field in the canonical space. Additionally, a neural point displacement field is introduced to enhance the capability of the blend weight field on modeling detailed human motions. Experiments show that our approach significantly outperforms recent human modeling methods.

Recent advances in implicit neural representations make it possible to generate free-viewpoint videos of the human from sparse view images. To avoid the expensive training for each person, previous methods adopt the generalizable human model and demonstrate impressive results. However, these methods usually rely on limited multi-view images typically collected in the studio or commercial high-quality 3D scans for training, which heavily prohibits their generalization capability for in-the-wild images. To solve this problem, we propose a new approach to learn a generalizable human model from a new source of data, i.e., Internet videos. These videos capture various human appearances and poses and record the performers from abundant viewpoints. To exploit the Internet data, we present a video self-supervised pipeline to enforce the local appearance consistency of each body part over different frames of the same video. Once learned, the human model enables realistic novel view synthesis from a single input image. Experiments show that our method can generate high-quality view synthesis on in-the-wild images while only training on monocular videos.

3D representation disentanglement aims to identify, decompose, and manipulate the underlying explanatory factors of 3D data, which helps AI fundamentally understand our 3D world. This task is currently under-explored and poses great challenges: (i) the 3D representations are complex and in general contains much more information than 2D image; (ii) many 3D representations are not well suited for gradient-based optimization, let alone disentanglement. To address these challenges, we use NeRF as a differentiable 3D representation, and introduce a self-supervised Navigation to identify interpretable semantic directions in the latent space. To our best knowledge, this novel method, dubbed NaviNeRF, is the first work to achieve fine-grained 3D disentanglement without any priors or supervisions. Specifically, NaviNeRF is built upon the generative NeRF pipeline, and equipped with an Outer Navigation Branch and an Inner Refinement Branch. They are complementary —— the outer navigation is to identify global-view semantic directions, and the inner refinement dedicates to fine-grained attributes. A synergistic loss is further devised to coordinate two branches. Extensive experiments demonstrate that NaviNeRF has a superior fine-grained 3D disentanglement ability than the previous 3D-aware models. Its performance is also comparable to editing-oriented models relying on semantic or geometry priors.*

Recent advances in implicit neural representations make it possible to reconstruct a human-body model from a monocular self-rotation video. While previous works present impressive results of human body reconstruction, the quality of reconstructed face and hands are relatively low. The main reason is that the image region occupied by these parts is very small compared to the body. To solve this problem, we propose a new approach named TotalSelfScan, which reconstructs the full-body model from several monocular self-rotation videos that focus on the face, hands, and body, respectively. Compared to recording a single video, this setting has almost no additional cost but provides more details of essential parts. To learn the full-body model, instead of encoding the whole body in a single network, we propose a multi-part representation to model separate parts and then fuse the part-specific observations into a single unified human model. Once learned, the full-body model enables rendering photorealistic free-viewpoint videos under novel human poses. Experiments show that TotalSelfScan can significantly improve the reconstruction and rendering quality on the face and hands compared to the existing methods. The code is available at \url{https://zju3dv.github.io/TotalSelfScan}.

This paper addresses the problem of reconstructing 3D poses of multiple people from a few calibrated camera views. The main challenge of this problem is to find the cross-view correspondences among noisy and incomplete 2D pose predictions. Most previous methods address this challenge by directly reasoning in 3D using a pictorial structure model, which is inefficient due to the huge state space. We propose a fast and robust approach to solve this problem. Our key idea is to use a multi-way matching algorithm to cluster the detected 2D poses in all views. Each resulting cluster encodes 2D poses of the same person across different views and consistent correspondences across the keypoints, from which the 3D pose of each person can be effectively inferred. The proposed convex optimization based multi-way matching algorithm is efficient and robust against missing and false detections, without knowing the number of people in the scene. Moreover, we propose to combine geometric and appearance cues for cross-view matching. Finally, an efficient tracking method is proposed to track the detected 3D poses across the multi-view video. The proposed approach achieves the state-of-the-art performance on the Campus and Shelf datasets, while being efficient for real-time applications.

In this paper, we introduce the new task of reconstructing 3D human pose from a single image in which we can see the person and the person’s image through a mirror. Compared to general scenarios of 3D pose estimation from a single view, the mirror reflection provides an additional view for resolving the depth ambiguity. We develop an optimization-based approach that exploits mirror symmetry constraints for accurate 3D pose reconstruction. We also provide a method to estimate the surface normal of the mirror from vanishing points in the single image. To validate the proposed approach, we collect a large-scale dataset named Mirrored-Human, which covers a large variety of human subjects, poses and backgrounds. The experiments demonstrate that, when trained on Mirrored-Human with our reconstructed 3D poses as pseudo ground-truth, the accuracy and generalizability of existing single-view 3D pose estimators can be largely improved. The code and dataset are available at https://zju3dv.github.io/Mirrored-Human/.

This paper addresses the challenge of reconstructing an animatable human model from a multi-view video. Some recent works have proposed to decompose a non-rigidly deforming scene into a canonical neural radiance field and a set of deformation fields that map observation-space points to the canonical space, thereby enabling them to learn the dynamic scene from images. However, they represent the deformation field as translational vector field or SE(3) field, which makes the optimization highly under-constrained. Moreover, these representations cannot be explicitly controlled by input motions. Instead, we introduce neural blend weight fields to produce the deformation fields. Based on the skeleton-driven deformation, blend weight fields are used with 3D human skeletons to generate observation-to-canonical and canonical-to-observation correspondences. Since 3D human skeletons are more observable, they can regularize the learning of deformation fields. Moreover, the learned blend weight fields can be combined with input skeletal motions to generate new deformation fields to animate the human model. Experiments show that our approach significantly outperforms recent human synthesis methods. The code and supplementary materials are available at https://zju3dv.github.io/animatable_nerf/.

Recovering human pose from RGB images and videos has drawn increasing attention in recent years owing to minimum sensor requirements and applicability in diverse fields such as human-computer interaction, robotics, video analytics, and augmented reality. Although a large amount of work has been devoted to this field, 3D human pose estimation based on monocular images or videos remains a very challenging task due to a variety of difficulties such as depth ambiguities, occlusion, background clutters, and lack of training data. In this survey, we summarize recent advances in monocular 3D human pose estimation. We provide a general taxonomy to cover existing approaches and analyze their capabilities and limitations. We also present a summary of extensively used datasets and metrics, and provide a quantitative comparison of some representative methods. Finally, we conclude with a discussion on realistic challenges and open problems for future research directions.

This paper addresses the problem of 3D pose estimation for multiple people in a few calibrated camera views. The main challenge of this problem is to find the cross-view correspondences among noisy and incomplete 2D pose predictions. Most previous methods address this challenge by directly reasoning in 3D using a pictorial structure model, which is inefficient due to the huge state space. We propose a fast and robust approach to solve this problem. Our key idea is to use a multi-way matching algorithm to cluster the detected 2D poses in all views. Each resulting cluster encodes 2D poses of the same person across different views and consistent correspondences across the keypoints, from which the 3D pose of each person can be effectively inferred. The proposed convex optimization based multi-way matching algorithm is efficient and robust against missing and false detections, without knowing the number of people in the scene. Moreover, we propose to combine geometric and appearance cues for cross-view matching. The proposed approach achieves significant performance gains from the state-of-the-art (96.3% vs. 90.6% and 96.9% vs. 88% on the Campus and Shelf datasets, respectively), while being efficient for real-time applications.