arXiv subjects

Zijian He

Publications and source records attributed to Zijian He.

At least 19 recordsLinked to original sources

arXiv2026

PAI-Actor: Cinematic Multi-Character Replacement in Dynamic Scenes

We present PAI-Actor, a cinematic multi-character animation framework for character replacement in dynamic movie scenes. Unlike conventional animation systems that mainly drive a single static image or a single subject, our goal is to replace and animate multiple characters within real video clips while preserving the original scene dynamics, camera motion, and background content. This setting is particularly challenging because the generated characters must remain consistent with the source performance in motion and interaction, while also matching the surrounding background in lighting, shadow, composition, and overall cinematic appearance. To address this, we formulate multi-character animation as a structure-guided human recovery problem and build a movie-driven training pipeline from high-quality film data. Furthermore, to support practical cinematic production, we introduce a bidirectional-to-autoregressive distillation framework: we first train a bidirectional diffusion transformer for high-quality short-clip generation at 1080P resolution, and then distill it into an autoregressive video-to-video model for efficient inference and longer video generation. Experiments show that PAI-Actor enables high-fidelity multi-character animation with strong scene consistency, cinematic visual quality, and efficient long-form generation.

cs.CV↗

arXiv2026

SpatialDiff: 3D-Aware Object Movement via Implicit Spatial Modeling

Recent advances in image editing allow impressive manipulation of objects, existing methods still struggle to handle spatial movement in complex scenes, such as objects span different depth layers or are partially occluded. Most image editing methods focus solely on prior information from 2D datasets, emphasizing planar features while lacking support for spatial structures. Even approaches that incorporate explicit positional information fail to capture true 3D spatial relationships, thus limiting accurate object movement in complex scenes. In this paper, we present SpatialDiff, a method that effectively captures 3D spatial structures, enabling precise and consistent object movements in complex scenes. Our core innovations are twofold: (1) Implicit 3D Spatial Modeling, which introduces 3D prior knowledge and enables the model to internally build a comprehensive understanding of the three-dimensional spatial structure; and (2) Global Spatial Supervision, which constrains the latent spatial features to enable the model to perceive changes in object spatial positions caused by editing operations. Experimental results demonstrate that our method significantly improves the accuracy and fidelity of spatial movement in complex scenes.

cs.CV↗

arXiv2026

LC-GRPO: Bridging Train-Inference Gap for Flow-Based GRPO with Langevin Correction

Flow-based generative models are typically sampled by solving a deterministic ordinary differential equation (ODE), whereas online reinforcement learning requires stochastic rollouts for policy exploration and optimization. Existing GRPO methods for flow models therefore replace the inference-time ODE with a stochastic differential equation (SDE) during training. Although the ODE and SDE share the same marginal distributions in continuous time, their finite-step discretizations can differ substantially. In particular, SDE rollouts often become blurry as the exploration noise increases, creating a mismatch between the samples used for reinforcement learning and those generated by the test-time ODE sampler. We introduce LC-GRPO, a flow-based GRPO framework with Langevin correction. Each rollout transition first takes an inference-aligned ODE Euler step and then applies a stochastic Langevin correction targeting the marginal distribution at the resulting timestep. The required score is recovered directly from the flow velocity, requiring no additional score model, while the resulting transition remains an isotropic Gaussian with a tractable likelihood for policy optimization. We theoretically show that, under suitable conditions, one Langevin correction step reduces the Wasserstein error of an imperfect ODE Euler step. At a matched randomness level, we further show that the proposed transition can be more accurate than the standard Euler--Maruyama discretization of the reverse SDE. Experiments on SD3.5-Medium, FLUX.1-Dev, and HunyuanVideo demonstrate that LC-GRPO consistently improves reward optimization across text-to-image and text-to-video tasks, preserves generation quality, and substantially narrows the gap between stochastic training rollouts and deterministic test-time ODE inference.

cs.LG↗

arXiv2026

Electron Delocalization versus Emission Coherence of Quantum Dot Superlattices

Cooperative emission is a collective quantum optical process that requires macroscopic phase coherence among coupled emitters. Recent observations of cooperative emission in QD superlattices have renewed interest in how such coherence emerges in nanostructured solids. Meanwhile, theoretical studies have long discussed the relationship between electronic delocalization and coherence, particularly whether delocalized states necessarily give rise to cooperative emission. This study addresses this question through power-dependent steady-state PL and time-resolved PL decay measurements. The findings indicate that, although the quantum resonance peak exhibits delocalized excitonic characteristics, it shows no signatures of cooperative radiation. In particular, neither superlinear intensity scaling nor power-dependent emission delay was observed, indicating the absence of cooperative-radiation signatures. This can be understood from two disorder-related aspects. Temperature-dependent spectroscopy reveals pronounced inhomogeneous broadening and low-temperature dark-exciton participation, pointing to intra-domain static disorder and exciton-state mixing. These effects collectively hinder the establishment of macroscopic coherence. The temperature dependence of the quantum resonance peak decay lifetime is consistent with two-dimensional exciton dynamics. This work provides direct experimental evidence that electronic delocalization can be decoupled from cooperative coherence in CdSe quantum dot superlattices.

cond-mat.mes-hall↗

arXiv2026

ROSE: Benchmarking the Perception-to-Action Gap in Multimodal Models

Multimodal large language models (MLLMs) are increasingly expected to act on visual information, yet the same scene may require different actions under different task contexts. How reliably can a model turn the same visual evidence into the action required by the current context? To answer this question, we introduce \textsc{ROSE} (\textbf{R}eference-conditioned \textbf{O}ddity and \textbf{S}ymbolic \textbf{E}xecution), a controlled benchmark that holds the visual scene fixed while varying region constraints and required symbolic outputs. Through coupled counting and coordinate-action tasks, \textsc{ROSE} tests whether models can infer an implicit majority reference and act on the resulting fine-grained visual evidence under changing contexts. Across nine recent MLLMs, performance drops by as much as 44.5 percentage points from counting-oriented tasks to region-conditioned action, despite 98.8\% human performance. The gap persists on paired scenes and regions for which the same model returns the correct count, while global-click and matched local controls show that coordinate grounding explains only part of the loss, revealing a distinct, model-dependent bottleneck in turning shared visual evidence into context-specific actions.

cs.CV↗

arXiv2026

Proprioceptive Invariant State Estimation for Humanoid Robots on Non-Inertial Ground

This paper presents an invariant extended Kalman filtering (InEKF) approach for real-time state estimation of humanoid robots operating on non-inertial ground using only onboard proprioceptive sensing. The proposed approach estimates the robot's base position and velocity relative to the moving ground frame without requiring direct measurements of ground motion or externally mounted sensors. By exploiting kinematic constraints at the stance foot through foot-mounted IMUs, the filter accounts for ground-induced nonlinearities in the process and measurement models while remaining fully proprioceptive. The estimator is formulated to admit a right-invariant measurement model, enabling favorable error dynamics under large initial uncertainties. Observability analysis establishes conditions under which the robot's relative base position and velocity are observable with respect to the non-inertial ground frame. Experiments with the Digit humanoid robot standing and squatting atop a swaying and pitching ground showcase a 96% speedup in convergence rate and an 80% reduction in position estimate errors over existing InEKFs. Walking experiments on a uni-axially rotating ground achieve an average estimation error of less than 9 cm for an initial error of up to 1 m.

cs.RO↗

arXiv2026

PAI-Studio: Cinematic Video Background Replacement with Camera-Aware Motion

We present PAI-Studio, a new reference-conditioned video synthesis task that addresses a long-standing challenge in cinematic background replacement: generating dynamic backgrounds aligned with foreground motion while preserving foreground identity, matching reference scene appearance, and achieving globally consistent illumination with realistic foreground relighting. Existing open-source systems and commercial APIs cannot simultaneously ensure motion-consistent background generation, high-fidelity foreground relighting and foreground identity preservation, often resulting in static backgrounds, inconsistent boundaries, and noticeable compositing artifacts. To bridge this gap, we build upon a Diffusion Transformer video backbone and reformulate the problem as an in-context conditional generation task. Through bidirectional attention, our model jointly captures foreground dynamics and background reference information within a unified architecture. We further construct a 30K-scale dataset sourced from high-quality films and online videos to support this task. Extensive evaluations demonstrate that our method significantly outperforms existing open-source and commercial API solutions.

cs.CV↗

arXiv2026

Nitsum: Serving Tiered LLM Requests with Adaptive Tensor Parallelism

LLM serving is increasingly multi-tenant: the same deployment must handle latency-critical interactive requests and more relaxed background workloads under a fixed GPU budget. This creates a tiered-SLO setting where maximizing overall goodput (requests that satisfy both TTFT and TPOT targets) is challenging because workload mix, request lengths, and load intensity vary over time. Existing systems mainly optimize request-level controls (e.g., queuing and batching) while keeping execution configuration largely static, which limits adaptation under multi-tier contention. We present Nitsum, a distributed LLM serving system that treats tensor parallelism (TP) as a first-class runtime control surface rather than a static deployment choice. Nitsum jointly optimizes TP level, prefill/decode GPU split, and request scheduling. To make frequent TP adaptation practical, Nitsum introduces TP-aware weight reuse and fast KV migration. Experiments on real traces and targeted microbenchmarks show that Nitsum improves SLO-compliant goodput over SoTA by up to 5.3 times.

cs.DC↗

arXiv2026

VDCores: Resource Decoupled Programming and Execution for Asynchronous GPU

Modern GPUs increasingly rely on specialized and asynchronous hardware units to deliver high performance. Yet these units are often underutilized because today's GPU software stacks still organize programming and execution around a monolithic kernel model that mismatches asynchronous hardware. To address this issue, Virtual Decoupled Engines (VDCores) presents a new decoupled programming and execution model for asynchronous GPUs. VDCores abstracts asynchronous hardware execution units as resource isolated virtual cores and represents workloads as dependency-connected micro-operations (micro-ops). this abstraction removes static orchestration from the programmer, enables automatic overlap of memory and compute based on dependency and resource readiness, and thereby improves utilization of asynchronous hardware resources. Realizing such a decoupled abstraction efficiently on today's GPUs is itself challenging, VDCores addresses this through a GPU-specialized programming model and GPU runtime design that preserves the flexibility while minimizing implementation overhead. Across four LLM inference workloads on GH200, H100, and RTX 6000 Pro GPUs, VDCores significantly improves decoding throughput by 24% on average and by up to 77% under dynamic inputs, while reducing kernel programming and specialization effort by 90%. We have open sourced VDCores at https://github.com/vdcores/vdcores.

cs.DC↗

arXiv2026

ChunQiuTR: Time-Keyed Temporal Retrieval in Classical Chinese Annals

Retrieval shapes how language models access and ground knowledge in retrieval-augmented generation (RAG). In historical research, the target is often not an arbitrary relevant passage, but the exact record for a specific regnal month, where temporal consistency matters as much as topical relevance. This is especially challenging for Classical Chinese annals, where time is expressed through terse, implicit, non-Gregorian reign phrases that must be interpreted from surrounding context, so semantically plausible evidence can still be temporally invalid. We introduce \textbf{ChunQiuTR}, a time-keyed retrieval benchmark built from the \textit{Spring and Autumn Annals} and its exegetical tradition. ChunQiuTR organizes records by month-level reign keys and includes chrono-near confounders that mirror realistic retrieval failures. We further propose \textbf{CTD} (Calendrical Temporal Dual-encoder), a time-aware dual-encoder that combines Fourier-based absolute calendrical context with relative offset biasing. Experiments show consistent gains over strong semantic dual-encoder baselines under time-keyed evaluation, supporting retrieval-time temporal consistency as a key prerequisite for faithful downstream historical RAG. Our code and datasets are available at \href{https://github.com/xbdxwyh/ChunQiuTR}{\texttt{github.com/xbdxwyh/ChunQiuTR}}.

cs.CL↗

arXiv2026

SceneExpander: Text-Guided 3D Scene Expansion via Free-Form View Insertion

World building with 3D scene representations is increasingly important for content creation, simulation, and interactive experiences, yet real workflows are inherently iterative: creators repeatedly extend existing scenes under user control. Motivated by this gap, we study text-guided 3D scene expansion via free-form view insertion. Starting from a real scene captured by multi-view images, a user specifies a text expansion intent, which a generative model materializes as an inserted view extending the scene coverage. Unlike simple object editing or style transfer within a fixed scene, the inserted view may be 3D-misaligned with the original reconstruction, introducing geometric shifts, hallucinated content, or view-dependent artifacts that disrupt global multi-view consistency. To address this challenge, we propose SceneExpander, which applies test-time adaptation to a parametric feed-forward 3D reconstruction model with two complementary distillation signals: anchor distillation stabilizes the captured scene using geometric cues from the captured views, while inserted-view self-distillation retains insertion-supported predictions to accommodate the misaligned view. Experiments on ETH scenes and online data demonstrate improved expansion behavior and reconstruction quality under misalignment.

cs.CV↗

arXiv2025

Direct Mapping of Intrinsic Topology of Bound States in the Continuum via Nonlinear Emission

The direct mapping of the intrinsic topology in a leaky photonic band is crucial and challenging in topological photonics. For instance, observables in bound states in the continuum (BICs) feature complex topological textures such as a polarization vortex in momentum space, which nonetheless is difficult to be characterized in far-field scattering, especially considering the dominant direct channel. Here, we propose and experimentally demonstrate a hybrid nonlinear metasurface that enables a direct visualization of the intrinsic topology in BICs via second-harmonic generation (SHG). The enhanced local-source of SHG from the ultrathin indium tin oxide can effectively excite the emissions from the eigenmodes of a TiO2 photonics crystal slab, achieving three-order enhancement of SHG magnitudes. Importantly, these enhanced SH emissions carry topological polarization textures of BICs to the far field. With this, we can directly construct polarization vector maps of symmetry-protected BICs and chiral symmetry-broken quasi-BICs, clearly visualizing the winding structure around V points, the generation and evolution of chiral C points. This work provides a universal approach for characterizing topological photonic systems via coherent nonlinearity processes, opening new avenues for studying topological phenomena in non-Hermitian photonic systems.

physics.optics↗

arXiv2025

Improving Chain-of-Thought Efficiency for Autoregressive Image Generation

Autoregressive multimodal large language models have recently gained popularity for image generation, driven by advances in foundation models. To enhance alignment and detail, newer approaches employ chain-of-thought (CoT) reasoning, expanding user inputs into elaborated prompts prior to image synthesis. However, this strategy can introduce unnecessary redundancy -- a phenomenon we call visual overthinking -- which increases computational costs and can introduce details that contradict the original prompt. In this work, we explore how to generate more concise CoT sequences for more efficient image generation. We introduce ShortCoTI, a lightweight optimization framework that encourages more concise CoT while preserving output image quality. ShortCoTI rewards more concise prompts with an adaptive function that scales according to an estimated difficulty for each task. Incorporating this reward into a reinforcement learning paradigm reduces prompt reasoning length by 54% while maintaining or slightly improving quality metrics across multiple benchmarks (T2I-CompBench, GenEval). Qualitative analysis shows that our method eliminates verbose explanations and repetitive refinements, producing reasoning prompts that are both concise and semantically rich. As a result, ShortCoTI improves computational efficiency without compromising the fidelity or visual appeal of generated images.

cs.CV↗

arXiv2025

PACE: Physics Augmentation for Coordinated End-to-end Reinforcement Learning toward Versatile Humanoid Table Tennis

Humanoid table tennis (TT) demands rapid perception, proactive whole-body motion, and agile footwork under strict timing--capabilities that remain difficult for end-to-end control policies. We propose a reinforcement learning (RL) framework that maps ball-position observations directly to whole-body joint commands for both arm striking and leg locomotion, strengthened by predictive signals and dense, physics-guided rewards. A lightweight learned predictor, fed with recent ball positions, estimates future ball states and augments the policy's observations for proactive decision-making. During training, a physics-based predictor supplies precise future states to construct dense, informative rewards that lead to effective exploration. The resulting policy attains strong performance across varied serve ranges (hit rate$\geq$96% and success rate$\geq$92%) in simulations. Ablation studies confirm that both the learned predictor and the predictive reward design are critical for end-to-end learning. Deployed zero-shot on a physical Booster T1 humanoid with 23 revolute joints, the policy produces coordinated lateral and forward-backward footwork with accurate, fast returns, suggesting a practical path toward versatile, competitive humanoid TT. We have open-sourced our RL training code at: https://github.com/purdue-tracelab/TTRL-ICRA2026

cs.RO↗

arXiv2025

One Model for All: Unified Try-On and Try-Off in Any Pose via LLM-Inspired Bidirectional Tweedie Diffusion

Recent diffusion-based approaches have made significant advances in image-based virtual try-on, enabling more realistic and end-to-end garment synthesis. However, most existing methods remain constrained by their reliance on exhibition garments and segmentation masks, as well as their limited ability to handle flexible pose variations. These limitations reduce their practicality in real-world scenarios; for instance, users cannot easily transfer garments worn by one person onto another, and the generated try-on results are typically restricted to the same pose as the reference image. In this paper, we introduce OMFA (One Model For All), a unified diffusion framework for both virtual try-on and try-off that operates without the need for exhibition garments and supports arbitrary poses. OMFA is inspired by the mask-based paradigm of discrete diffusion language models and unifies try-on and try-off within a bidirectional framework. It is built upon a Bidirectional Tweedie Diffusion process for target-selective denoising in latent space. Instead of imposing lower body constraints, OMFA is an entirely mask-free framework that requires only a single portrait and a target garment as inputs, and is designed to support flexible outfit combinations and cross-person garment transfer, making it better aligned with practical usage scenarios. Additionally, by leveraging SMPL-X-based pose conditioning, OMFA supports multi-view and arbitrary-pose try-on from just one image. Extensive experiments demonstrate that OMFA achieves state-of-the-art results on both try-on and try-off tasks, providing a practical and generalizable solution for virtual garment synthesis. Project page: https://onemodelforall.github.io

cs.CV↗

arXiv2025

Near-Field Propagation and Spatial Non-Stationarity Channel Model for 6-24 GHz (FR3) Extremely Large-Scale MIMO: Adopted by 3GPP for 6G

Next generation cellular deployments are expected to exploit the 6-24 GHz frequency range 3 (FR3) and extremely large-scale multiple-input multiple-output (XL-MIMO) to enable ultra-high data rates and reliability. However, the significantly enlarged antenna apertures and higher carrier frequencies render the far-field and spatial stationarity assumptions in the existing 3rd generation partnership project (3GPP) channel models invalid, giving rise to new features such as near-field propagation and spatial non-stationarity (SNS). Despite extensive prior research, incorporating these new features within the standardized channel modeling framework remains an open issue. To address this, this paper presents a channel modeling framework for XL-MIMO systems that incorporates both near-field and SNS features, adopted by 3GPP. For the near-field propagation feature, the framework models the distances from the base station (BS) and user equipment to the spherical-wave sources associated with clusters. These distances are used to characterize element-wise variations of path parameters, such as nonlinear changes in phase and angle. To capture the effect of SNS at the BS side, a stochastic-based approach is proposed to model SNS caused by incomplete scattering, by establishing power attenuation factors from visibility probability and visibility region to characterize antenna element-wise path power variation. In addition, a physical blocker-based approach is introduced to model SNS effects caused by partial blockage. Finally, a simulation framework for near-field and SNS is developed within the structure of the existing 3GPP channel model. Performance evaluations demonstrate that the near-field model captures higher channel capacity potential compared to the far-field model. Coupling loss results indicate that SNS leads to more pronounced propagation fading relative to the spatial stationary model.

eess.SP↗

arXiv2025

HMVLM: Multistage Reasoning-Enhanced Vision-Language Model for Long-Tailed Driving Scenarios

We present HaoMo Vision-Language Model (HMVLM), an end-to-end driving framework that implements the slow branch of a cognitively inspired fast-slow architecture. A fast controller outputs low-level steering, throttle, and brake commands, while a slow planner-a large vision-language model-generates high-level intents such as "yield to pedestrian" or "merge after the truck" without compromising latency. HMVLM introduces three upgrades: (1) selective five-view prompting with an embedded 4s history of ego kinematics, (2) multi-stage chain-of-thought (CoT) prompting that enforces a Scene Understanding -> Driving Decision -> Trajectory Inference reasoning flow, and (3) spline-based trajectory post-processing that removes late-stage jitter and sharp turns. Trained on the Waymo Open Dataset, these upgrades enable HMVLM to achieve a Rater Feedback Score (RFS) of 7.7367, securing 2nd place in the 2025 Waymo Vision-based End-to-End (E2E) Driving Challenge and surpassing the public baseline by 2.77%.

cs.CV↗

arXiv2025

Mod-Adapter: Tuning-Free and Versatile Multi-concept Personalization via Modulation Adapter

Personalized text-to-image generation aims to synthesize images of user-provided concepts in diverse contexts. Despite recent progress in multi-concept personalization, most are limited to object concepts and struggle to customize abstract concepts (e.g., pose, lighting). Some methods have begun exploring multi-concept personalization supporting abstract concepts, but they require test-time fine-tuning for each new concept, which is time-consuming and prone to overfitting on limited training images. In this work, we propose a novel tuning-free method for multi-concept personalization that can effectively customize both object and abstract concepts without test-time fine-tuning. Our method builds upon the modulation mechanism in pre-trained Diffusion Transformers (DiTs) model, leveraging the localized and semantically meaningful properties of the modulation space. Specifically, we propose a novel module, Mod-Adapter, to predict concept-specific modulation direction for the modulation process of concept-related text tokens. It introduces vision-language cross-attention for extracting concept visual features, and Mixture-of-Experts (MoE) layers that adaptively map the concept features into the modulation space. Furthermore, to mitigate the training difficulty caused by the large gap between the concept image space and the modulation space, we introduce a VLM-guided pre-training strategy that leverages the strong image understanding capabilities of vision-language models to provide semantic supervision signals. For a comprehensive comparison, we extend a standard benchmark by incorporating abstract concepts. Our method achieves state-of-the-art performance in multi-concept personalization, supported by quantitative, qualitative, and human evaluations.

cs.CV↗