Jiale Xu

Jiale Xu, Wang Zhao, Ying Shan

Autoregressive mesh generation has gained attention by tokenizing meshes into sequences and training models in a language-modeling fashion. However, existing approaches suffer from two fundamental limitations: (i) low tokenization efficiency, which yields long token sequences and prevents scaling to high-poly meshes, and (ii) absence of geometry-aware guidance, as generation is conditioned only on global shape embeddings rather than local surface cues. We introduce MeshWeaver, an autoregressive framework that treats mesh generation as a surface weaving process by directly predicting the next vertex instead of independent coordinates. At its core is a multi-level sparse-voxel encoder that injects geometric context into the generative process in three complementary ways: providing voxel features as vertex representations, guiding token prediction via cross-attention to voxel features, and serving as a structural scaffold that constrains generation around the input surface. Our hierarchical design enables coarse-to-fine vertex prediction in a single decoding step, while tightly coupling the generative model with 3D geometry. Extensive experiments demonstrate that MeshWeaver achieves a state-of-the-art compression ratio of 18%, can generate meshes with up to 16K faces, and significantly improves geometric fidelity over prior approaches.

Jiale Xu, Xintao Wang, Weihao Cheng et al.

Recent CLIP-guided 3D optimization methods, such as DreamFields and PureCLIPNeRF, have achieved impressive results in zero-shot text-to-3D synthesis. However, due to scratch training and random initialization without prior knowledge, these methods often fail to generate accurate and faithful 3D structures that conform to the input text. In this paper, we make the first attempt to introduce explicit 3D shape priors into the CLIP-guided 3D optimization process. Specifically, we first generate a high-quality 3D shape from the input text in the text-to-shape stage as a 3D shape prior. We then use it as the initialization of a neural radiance field and optimize it with the full prompt. To address the challenging text-to-shape generation task, we present a simple yet effective approach that directly bridges the text and image modalities with a powerful text-to-image diffusion model. To narrow the style domain gap between the images synthesized by the text-to-image diffusion model and shape renderings used to train the image-to-shape generator, we further propose to jointly optimize a learnable text prompt and fine-tune the text-to-image diffusion model for rendering-style image generation. Our method, Dream3D, is capable of generating imaginative 3D content with superior visual quality and shape accuracy compared to state-of-the-art methods.

Ziming Zhong, Yanxu Xu, Jing Li et al.

We present MeshSegmenter, a simple yet effective framework designed for zero-shot 3D semantic segmentation. This model successfully extends the powerful capabilities of 2D segmentation models to 3D meshes, delivering accurate 3D segmentation across diverse meshes and segment descriptions. Specifically, our model leverages the Segment Anything Model (SAM) model to segment the target regions from images rendered from the 3D shape. In light of the importance of the texture for segmentation, we also leverage the pretrained stable diffusion model to generate images with textures from 3D shape, and leverage SAM to segment the target regions from images with textures. Textures supplement the shape for segmentation and facilitate accurate 3D segmentation even in geometrically non-prominent areas, such as segmenting a car door within a car mesh. To achieve the 3D segments, we render 2D images from different views and conduct segmentation for both textured and untextured images. Lastly, we develop a multi-view revoting scheme that integrates 2D segmentation results and confidence scores from various views onto the 3D mesh, ensuring the 3D consistency of segmentation results and eliminating inaccuracies from specific perspectives. Through these innovations, MeshSegmenter offers stable and reliable 3D segmentation results both quantitatively and qualitatively, highlighting its potential as a transformative tool in the field of 3D zero-shot segmentation. The code is available at \url{https://github.com/zimingzhong/MeshSegmenter}.

Jiale Xu, Rui Zhang, Yi Xiong et al.

Large Language Models are increasingly being deployed in datacenters. Serving these models requires careful memory management, as their memory usage includes static weights, dynamic activations, and key-value caches. While static weights are constant and predictable, dynamic components such as activations and KV caches change frequently during runtime, presenting significant challenges for efficient memory management. Modern LLM serving systems typically handle runtime memory and KV caches at distinct abstraction levels: runtime memory management relies on static tensor abstractions, whereas KV caches utilize a page table-based virtualization layer built on top of the tensor abstraction. This virtualization dynamically manages KV caches to mitigate memory fragmentation. However, this dual-level approach fundamentally isolates runtime memory and KV cache management, resulting in suboptimal memory utilization under dynamic workloads, which can lead to a nearly 20% drop in throughput. To address these limitations, we propose eLLM, an elastic memory management framework inspired by the classical memory ballooning mechanism in operating systems. The core components of eLLM include: (1) Virtual Tensor Abstraction, which decouples the virtual address space of tensors from the physical GPU memory, creating a unified and flexible memory pool; (2) an Elastic Memory Mechanism that dynamically adjusts memory allocation through runtime memory inflation and deflation, leveraging CPU memory as an extensible buffer; and (3) a Lightweight Scheduling Strategy employing SLO-aware policies to optimize memory utilization and effectively balance performance trade-offs under stringent SLO constraints. Comprehensive evaluations demonstrate that eLLM significantly outperforms state-of-the-art systems, 2.32x higher decoding throughput, and supporting 3x larger batch sizes for 128K-token inputs.

Jiale Xu, Xintao Wang, Yan-Pei Cao et al.

Enhancing AI systems to perform tasks following human instructions can significantly boost productivity. In this paper, we present InstructP2P, an end-to-end framework for 3D shape editing on point clouds, guided by high-level textual instructions. InstructP2P extends the capabilities of existing methods by synergizing the strengths of a text-conditioned point cloud diffusion model, Point-E, and powerful language models, enabling color and geometry editing using language instructions. To train InstructP2P, we introduce a new shape editing dataset, constructed by integrating a shape segmentation dataset, off-the-shelf shape programs, and diverse edit instructions generated by a large language model, ChatGPT. Our proposed method allows for editing both color and geometry of specific regions in a single forward pass, while leaving other regions unaffected. In our experiments, InstructP2P shows generalization capabilities, adapting to novel shape categories and instructions, despite being trained on a limited amount of data.

Shenhan Qian, Jiale Xu, Ziwei Liu et al.

We propose united implicit functions (UNIF), a part-based method for clothed human reconstruction and animation with raw scans and skeletons as the input. Previous part-based methods for human reconstruction rely on ground-truth part labels from SMPL and thus are limited to minimal-clothed humans. In contrast, our method learns to separate parts from body motions instead of part supervision, thus can be extended to clothed humans and other articulated objects. Our Partition-from-Motion is achieved by a bone-centered initialization, a bone limit loss, and a section normal loss that ensure stable part division even when the training poses are limited. We also present a minimal perimeter loss for SDF to suppress extra surfaces and part overlapping. Another core of our method is an adjacent part seaming algorithm that produces non-rigid deformations to maintain the connection between parts which significantly relieves the part-based artifacts. Under this algorithm, we further propose "Competing Parts", a method that defines blending weights by the relative position of a point to bones instead of the absolute position, avoiding the generalization problem of neural implicit functions with inverse LBS (linear blend skinning). We demonstrate the effectiveness of our method by clothed human body reconstruction and animation on the CAPE and the ClothSeq datasets.

Jiale Xu, Rui Zhang, Cong Guo et al.

Large Language Models (LLMs) are widely used across various domains, processing millions of daily requests. This surge in demand poses significant challenges in optimizing throughput and latency while keeping costs manageable. The Key-Value (KV) cache, a standard method for retaining previous computations, makes LLM inference highly bounded by memory. While batching strategies can enhance performance, they frequently lead to significant memory fragmentation. Even though cutting-edge systems like vLLM mitigate KV cache fragmentation using paged Attention mechanisms, they still suffer from inefficient memory and computational operations due to the tightly coupled page management and computation kernels. This study introduces the vTensor, an innovative tensor structure for LLM inference based on GPU virtual memory management (VMM). vTensor addresses existing limitations by decoupling computation from memory defragmentation and offering dynamic extensibility. Our framework employs a CPU-GPU heterogeneous approach, ensuring efficient, fragmentation-free memory management while accommodating various computation kernels across different LLM architectures. Experimental results indicate that vTensor achieves an average speedup of 1.86x across different models, with up to 2.42x in multi-turn chat scenarios. Additionally, vTensor provides average speedups of 2.12x and 3.15x in kernel evaluation, reaching up to 3.92x and 3.27x compared to SGLang Triton prefix-prefilling kernels and vLLM paged Attention kernel, respectively. Furthermore, it frees approximately 71.25% (57GB) of memory on the NVIDIA A100 GPU compared to vLLM, enabling more memory-intensive workloads.

Lin Jiang, Qingshan She, Jiale Xu et al.

Zero-shot EEG-to-image retrieval aims to decode perceived visual content from electroencephalography (EEG) by aligning neural responses with pretrained visual representations, providing a promising route toward scalable visual neural decoding and practical brain-computer interfaces. However, robust EEG-to-image retrieval remains challenging, because prior methods usually rely on either a single fixed visual target or a subject-invariant target construction scheme. Such designs overlook two important properties of visually evoked EEG signals: they preserve information across multiple representational scales, and the visual granularity best matched to EEG may vary across subjects. To address these issues, subject-aware multi-granularity alignment (SAMGA) framework is proposed for zero-shot EEG-to-image retrieval. SAMGA first constructs a subject-aware visual supervision target by adaptively aggregating multiple intermediate representations from a pretrained vision encoder, allowing the model to absorb subject-dependent granularity deviations during training while preserving subject-agnostic inference. Building on this adaptive target construction, a coarse-to-fine cross-modal alignment strategy is further designed with a shared encoder wherein the coarse stage stabilizes the shared semantic geometry and reduces subject-induced distribution shift, and the fine stage further improves instance-level retrieval discrimination. Extensive experiments on the THINGS-EEG benchmark demonstrate that the proposed method achieves 91.3% Top-1 and 98.8% Top-5 accuracy in the intra-subject setting, and 34.4% Top-1 and 64.8% Top-5 accuracy in the inter-subject setting, outperforming recent state-of-the-art methods.

Jiale Xu, Weihao Cheng, Yiming Gao et al.

We present InstantMesh, a feed-forward framework for instant 3D mesh generation from a single image, featuring state-of-the-art generation quality and significant training scalability. By synergizing the strengths of an off-the-shelf multiview diffusion model and a sparse-view reconstruction model based on the LRM architecture, InstantMesh is able to create diverse 3D assets within 10 seconds. To enhance the training efficiency and exploit more geometric supervisions, e.g, depths and normals, we integrate a differentiable iso-surface extraction module into our framework and directly optimize on the mesh representation. Experimental results on public datasets demonstrate that InstantMesh significantly outperforms other latest image-to-3D baselines, both qualitatively and quantitatively. We release all the code, weights, and demo of InstantMesh, with the intention that it can make substantial contributions to the community of 3D generative AI and empower both researchers and content creators.

Yaowei Li, Xiaoyu Li, Zhaoyang Zhang et al.

Image customization, a crucial technique for industrial media production, aims to generate content that is consistent with reference images. However, current approaches conventionally separate image customization into position-aware and position-free customization paradigms and lack a universal framework for diverse customization, limiting their applications across various scenarios. To overcome these limitations, we propose IC-Custom, a unified framework that seamlessly integrates position-aware and position-free image customization through in-context learning. IC-Custom concatenates reference images with target images to a polyptych, leveraging DiT's multi-modal attention mechanism for fine-grained token-level interactions. We propose the In-context Multi-Modal Attention (ICMA) mechanism, which employs learnable task-oriented register tokens and boundary-aware positional embeddings to enable the model to effectively handle diverse tasks and distinguish between inputs in polyptych configurations. To address the data gap, we curated a 12K identity-consistent dataset with 8K real-world and 4K high-quality synthetic samples, avoiding the overly glossy, oversaturated look typical of synthetic data. IC-Custom supports various industrial applications, including try-on, image insertion, and creative IP customization. Extensive evaluations on our proposed ProductBench and the publicly available DreamBench demonstrate that IC-Custom significantly outperforms community workflows, closed-source models, and state-of-the-art open-source approaches. IC-Custom achieves about 73\% higher human preference across identity consistency, harmony, and text alignment metrics, while training only 0.4\% of the original model parameters. Project page: https://liyaowei-stu.github.io/project/IC_Custom

Jiale Xu, Jia Zheng, Yanyu Xu et al.

Existing view synthesis methods mainly focus on the perspective images and have shown promising results. However, due to the limited field-of-view of the pinhole camera, the performance quickly degrades when large camera movements are adopted. In this paper, we make the first attempt to generate novel views from a single indoor panorama and take the large camera translations into consideration. To tackle this challenging problem, we first use Convolutional Neural Networks (CNNs) to extract the deep features and estimate the depth map from the source-view image. Then, we leverage the room layout prior, a strong structural constraint of the indoor scene, to guide the generation of target views. More concretely, we estimate the room layout in the source view and transform it into the target viewpoint as guidance. Meanwhile, we also constrain the room layout of the generated target-view images to enforce geometric consistency. To validate the effectiveness of our method, we further build a large-scale photo-realistic dataset containing both small and large camera translations. The experimental results on our challenging dataset demonstrate that our method achieves state-of-the-art performance. The project page is at https://github.com/bluestyle97/PNVS.

Qian Feng, Pengfei Li, Rongshan Gao et al.

Remote photoplethysmography (rPPG) based on traditional frame-based cameras often struggles with motion artifacts and limited temporal resolution. To address these limitations, we introduce EMPD (Event-based Multimodal Physiological Dataset), the first benchmark dataset specifically designed for non-contact physiological sensing via event cameras. The dataset leverages a laser-assisted acquisition system where a high-coherence laser modulates subtle skin vibrations from the radial artery into significant signals detectable by a neuromorphic sensor. The hardware platform integrates a high-resolution event camera to capture micro-motions and intensity transients, an industrial RGB camera to provide traditional rPPG benchmarks, and a clinical-grade pulse oximeter to record ground truth PPG waveforms. EMPD contains 193 valid records collected from 83 subjects, covering a wide heart rate range (40-110 BPM) under both resting and post-exercise conditions. By providing precisely synchronized multimodal data with microsecond-level temporal precision, EMPD serves as a crucial resource for developing robust algorithms in the field of neuromorphic physiological monitoring. The dataset is publicly available at: https://doi.org/10.5281/zenodo.18765701

Jiale Xu, Shenghua Gao, Ying Shan

Sparse-view reconstruction models typically require precise camera poses, yet obtaining these parameters from sparse-view images remains challenging. We introduce FreeSplatter, a scalable feed-forward framework that generates high-quality 3D Gaussians from uncalibrated sparse-view images while estimating camera parameters within seconds. Our approach employs a streamlined transformer architecture where self-attention blocks facilitate information exchange among multi-view image tokens, decoding them into pixel-aligned 3D Gaussian primitives within a unified reference frame. This representation enables both high-fidelity 3D modeling and efficient camera parameter estimation using off-the-shelf solvers. We develop two specialized variants--for object-centric and scene-level reconstruction--trained on comprehensive datasets. Remarkably, FreeSplatter outperforms several pose-dependent Large Reconstruction Models (LRMs) by a notable margin while achieving comparable or even better pose estimation accuracy compared to state-of-the-art pose-free reconstruction approach MASt3R in challenging benchmarks. Beyond technical benchmarks, FreeSplatter streamlines text/image-to-3D content creation pipelines, eliminating the complexity of camera pose management while delivering exceptional visual fidelity.

Minghao Yin, Wenbo Hu, Jiale Xu et al.

Recent breakthroughs in 3D generative modeling have yielded remarkable progress in static shape synthesis, yet high-fidelity dynamic 4D generation remains elusive, hindered by temporal artifacts and prohibitive computational demand. We present Sculpt4D, a native 4D generative framework that seamlessly integrates efficient temporal modeling into a pretrained 3D Diffusion Transformer (Hunyuan3D 2.1), thereby mitigating the scarcity of 4D training data. At its core lies a Block Sparse Attention mechanism that preserves object identity by anchoring to the initial frame while capturing rich motion dynamics via a time-decaying sparse mask. This design faithfully models complex spatiotemporal dependencies with high fidelity, while sidestepping the quadratic overhead of full attention and reducing network total computation by 56%. Consequently, Sculpt4D establishes a new state-of-the-art in temporally coherent 4D synthesis and charts a path toward efficient and scalable 4D generation.

Lingen Li, Zhaoyang Zhang, Yaowei Li et al.

Recent advancements in generative models have significantly improved novel view synthesis (NVS) from multi-view data. However, existing methods depend on external multi-view alignment processes, such as explicit pose estimation or pre-reconstruction, which limits their flexibility and accessibility, especially when alignment is unstable due to insufficient overlap or occlusions between views. In this paper, we propose NVComposer, a novel approach that eliminates the need for explicit external alignment. NVComposer enables the generative model to implicitly infer spatial and geometric relationships between multiple conditional views by introducing two key components: 1) an image-pose dual-stream diffusion model that simultaneously generates target novel views and condition camera poses, and 2) a geometry-aware feature alignment module that distills geometric priors from dense stereo models during training. Extensive experiments demonstrate that NVComposer achieves state-of-the-art performance in generative multi-view NVS tasks, removing the reliance on external alignment and thus improving model accessibility. Our approach shows substantial improvements in synthesis quality as the number of unposed input views increases, highlighting its potential for more flexible and accessible generative NVS systems. Our project page is available at https://lg-li.github.io/project/nvcomposer

Jinpeng Liu, Jiale Xu, Weihao Cheng et al.

We introduce NovelGS, a diffusion model for Gaussian Splatting (GS) given sparse-view images. Recent works leverage feed-forward networks to generate pixel-aligned Gaussians, which could be fast rendered. Unfortunately, the method was unable to produce satisfactory results for areas not covered by the input images due to the formulation of these methods. In contrast, we leverage the novel view denoising through a transformer-based network to generate 3D Gaussians. Specifically, by incorporating both conditional views and noisy target views, the network predicts pixel-aligned Gaussians for each view. During training, the rendered target and some additional views of the Gaussians are supervised. During inference, the target views are iteratively rendered and denoised from pure noise. Our approach demonstrates state-of-the-art performance in addressing the multi-view image reconstruction challenge. Due to generative modeling of unseen regions, NovelGS effectively reconstructs 3D objects with consistent and sharp textures. Experimental results on publicly available datasets indicate that NovelGS substantially surpasses existing image-to-3D frameworks, both qualitatively and quantitatively. We also demonstrate the potential of NovelGS in generative tasks, such as text-to-3D and image-to-3D, by integrating it with existing multiview diffusion models. We will make the code publicly accessible.

Wang Zhao, Yan-Pei Cao, Jiale Xu et al.

Procedural Content Generation (PCG) is powerful in creating high-quality 3D contents, yet controlling it to produce desired shapes is difficult and often requires extensive parameter tuning. Inverse Procedural Content Generation aims to automatically find the best parameters under the input condition. However, existing sampling-based and neural network-based methods still suffer from numerous sample iterations or limited controllability. In this work, we present DI-PCG, a novel and efficient method for Inverse PCG from general image conditions. At its core is a lightweight diffusion transformer model, where PCG parameters are directly treated as the denoising target and the observed images as conditions to control parameter generation. DI-PCG is efficient and effective. With only 7.6M network parameters and 30 GPU hours to train, it demonstrates superior performance in recovering parameters accurately, and generalizing well to in-the-wild images. Quantitative and qualitative experiment results validate the effectiveness of DI-PCG in inverse PCG and image-to-3D generation tasks. DI-PCG offers a promising approach for efficient inverse PCG and represents a valuable exploration step towards a 3D generation path that models how to construct a 3D asset using parametric models.

Kunhao Liu, Wenbo Hu, Jiale Xu et al.

Streaming video generation, as one fundamental component in interactive world models and neural game engines, aims to generate high-quality, low-latency, and temporally coherent long video streams. However, most existing work suffers from severe error accumulation that often significantly degrades the generated stream videos over long horizons. We design Rolling Forcing, a novel video generation technique that enables streaming long videos with minimal error accumulation. Rolling Forcing comes with three novel designs. First, instead of iteratively sampling individual frames, which accelerates error propagation, we design a joint denoising scheme that simultaneously denoises multiple frames with progressively increasing noise levels. This design relaxes the strict causality across adjacent frames, effectively suppressing error growth. Second, we introduce the attention sink mechanism into the long-horizon stream video generation task, which allows the model to keep key value states of initial frames as a global context anchor and thereby enhances long-term global consistency. Third, we design an efficient training algorithm that enables few-step distillation over largely extended denoising windows. This algorithm operates on non-overlapping windows and mitigates exposure bias conditioned on self-generated histories. Extensive experiments show that Rolling Forcing enables real-time streaming generation of multi-minute videos on a single GPU, with substantially reduced error accumulation.

Wang Zhao, Yan-Pei Cao, Jiale Xu et al.

We present Assembler, a scalable and generalizable framework for 3D part assembly that reconstructs complete objects from input part meshes and a reference image. Unlike prior approaches that mostly rely on deterministic part pose prediction and category-specific training, Assembler is designed to handle diverse, in-the-wild objects with varying part counts, geometries, and structures. It addresses the core challenges of scaling to general 3D part assembly through innovations in task formulation, representation, and data. First, Assembler casts part assembly as a generative problem and employs diffusion models to sample plausible configurations, effectively capturing ambiguities arising from symmetry, repeated parts, and multiple valid assemblies. Second, we introduce a novel shape-centric representation based on sparse anchor point clouds, enabling scalable generation in Euclidean space rather than SE(3) pose prediction. Third, we construct a large-scale dataset of over 320K diverse part-object assemblies using a synthesis and filtering pipeline built on existing 3D shape repositories. Assembler achieves state-of-the-art performance on PartNet and is the first to demonstrate high-quality assembly for complex, real-world objects. Based on Assembler, we further introduce an interesting part-aware 3D modeling system that generates high-resolution, editable objects from images, demonstrating potential for interactive and compositional design. Project page: https://assembler3d.github.io

Yue-Jiang Dong, Wang Zhao, Jiale Xu et al.

Diffusion-based video depth estimation methods have achieved remarkable success with strong generalization ability. However, predicting depth for long videos remains challenging. Existing methods typically split videos into overlapping sliding windows, leading to accumulated scale discrepancies across different windows, particularly as the number of windows increases. Additionally, these methods rely solely on 2D diffusion priors, overlooking the inherent 3D geometric structure of video depths, which results in geometrically inconsistent predictions. In this paper, we propose DepthSync, a novel, training-free framework using diffusion guidance to achieve scale- and geometry-consistent depth predictions for long videos. Specifically, we introduce scale guidance to synchronize the depth scale across windows and geometry guidance to enforce geometric alignment within windows based on the inherent 3D constraints in video depths. These two terms work synergistically, steering the denoising process toward consistent depth predictions. Experiments on various datasets validate the effectiveness of our method in producing depth estimates with improved scale and geometry consistency, particularly for long videos.

Yuting Xiao, Jiale Xu, Shenghua Gao

Benefiting from the continuous representation ability, deep implicit functions can represent a shape at infinite resolution. However, extracting high-resolution iso-surface from an implicit function requires forward-propagating a network with a large number of parameters for numerous query points, thus preventing the generation speed. Inspired by the Taylor series, we propose Taylo3DNet to accelerate the inference of implicit shape representations. Taylor3DNet exploits a set of discrete landmark points and their corresponding Taylor series coefficients to represent the implicit field of a 3D shape, and the number of landmark points is independent of the resolution of the iso-surface extraction. Once the coefficients corresponding to the landmark points are predicted, the network evaluation for each query point can be simplified as a low-order Taylor series calculation with several nearest landmark points. Based on this efficient representation, our Taylor3DNet achieves a significantly faster inference speed than classical network-based implicit functions. We evaluate our approach on reconstruction tasks with various input types, and the results demonstrate that our approach can improve the inference speed by a large margin without sacrificing the performance compared with state-of-the-art baselines.

Xianing Chen, Chunlin Xu, Qiong Cao et al.

Transformers have shown preferable performance on many vision tasks. However, for the task of person re-identification (ReID), vanilla transformers leave the rich contexts on high-order feature relations under-exploited and deteriorate local feature details, which are insufficient due to the dramatic variations of pedestrians. In this work, we propose an Omni-Relational High-Order Transformer (OH-Former) to model omni-relational features for ReID. First, to strengthen the capacity of visual representation, instead of obtaining the attention matrix based on pairs of queries and isolated keys at each spatial location, we take a step further to model high-order statistics information for the non-local mechanism. We share the attention weights in the corresponding layer of each order with a prior mixing mechanism to reduce the computation cost. Then, a convolution-based local relation perception module is proposed to extract the local relations and 2D position information. The experimental results of our model are superior promising, which show state-of-the-art performance on Market-1501, DukeMTMC, MSMT17 and Occluded-Duke datasets.