Pre-Trained Video Diffusion Models

Aligning streaming autoregressive (AR) video generators with human preferences is challenging. Existing reinforcement learning methods predominantly rely on noise-based exploration and SDE-based surrogate policies that are mismatched to the deterministic ODE dynamics of distilled AR models, and tend to perturb low-level appearance rather than the high-level semantic storyline progression critical for long-horizon coherence. To address these limitations, we present KVPO, an ODE-native online Group Relative Policy Optimization (GRPO) framework for aligning streaming video generators. For diversity exploration, KVPO introduces a causal-semantic exploration paradigm that relocates the source of variation from stochastic noise to the historical KV cache. By stochastically routing historical KV entries, it constructs semantically diverse generation branches that remain strictly on the data manifold. For policy modeling, KVPO introduces a velocity-field surrogate policy based on Trajectory Velocity Energy (TVE), which quantifies branch likelihood in flow-matching velocity space and yields a reward-weighted contrastive objective fully consistent with the native ODE formulation. Experiments on multiple distilled AR video generators demonstrate consistent gains in visual quality, motion quality, and text-video alignment across both single-prompt short-video and multi-prompt long-video settings.

Recent text-to-video diffusion transformers generate visually compelling frames, yet still struggle with temporal coherence, often producing flickering, drifting, or unstable motion. We show that these failures leave a clear imprint inside the model: incoherent videos consistently exhibit irregular, fragmented temporal diagonals in their intermediate self-attention maps, whereas stable motion corresponds to smooth, band-diagonal patterns. Building on this observation, we introduce TeDiO, a training-free, inference-time method that reinforces temporal consistency by regularizing these internal attention patterns. TeDiO estimates diagonal smoothness, identifies unstable regions, and performs lightweight latent updates that promote coherent frame-to-frame dynamics, without modifying model weights or using external motion supervision. Across multiple video diffusion models (e.g., Wan2.1, CogVideoX), TeDiO delivers markedly smoother motion while preserving per-frame visual quality, offering an efficient plug-and-play approach to improving dynamic realism in modern video generation systems.

Trajectory-controlled human motion generation aims to synthesize realistic human motions conditioned on both textual descriptions and spatial trajectories. However, existing methods suffer from two critical limitations: first, the conflict between text and trajectory conditions disrupts the denoising process, resulting in compromised motion quality or inaccurate trajectory following; second, the use of redundant motion representations introduces inconsistencies between motion components, leading to instability during trajectory control. To address these challenges, we propose CMC, a decoupled framework that effectively coordinates text and trajectory conditions through a divide-and-conquer strategy. CMC follows a divide-and-conquer paradigm, comprising two cascaded stages: Trajectory Control and Motion Completion. In the first stage, a diffusion model generates a simplified representation of the controlled joints under trajectory guidance, based on the given trajectories, ensuring accurate and stable trajectory following. In the second stage, a text-conditioned diffusion inpainting model generates full-body motions using the simplified representation from the first stage as partial observations. To mitigate overfitting caused by limited inpainting training data, we further introduce the Selective Inpainting Mechanism (SIM), which alternates between text-to-motion generation and motion inpainting tasks during training. Experiments on HumanML3D and KIT datasets demonstrate that CMC achieves state-of-the-art performance in control accuracy and motion quality, demonstrating its effectiveness in coordinating multimodal conditions and representations.

Few-step video generation has been significantly advanced by consistency distillation. However, the performance of consistency-distilled models often degrades as more sampling steps are allocated at test time, limiting their effectiveness for any-step video diffusion. This limitation arises because consistency distillation replaces the original probability-flow ODE trajectory with a consistency-sampling trajectory, weakening the desirable test-time scaling behavior of ODE sampling. To address this limitation, we introduce AnyFlow, the first any-step video diffusion distillation framework based on flow maps. Instead of distilling a model for only a few fixed sampling steps, AnyFlow optimizes the full ODE sampling trajectory. To this end, we shift the distillation target from endpoint consistency mapping $(z_{t}\rightarrow z_{0})$ to flow-map transition learning $(z_{t}\rightarrow z_{r})$ over arbitrary time intervals. We further propose Flow Map Backward Simulation, which decomposes a full Euler rollout into shortcut flow-map transitions, enabling efficient on-policy distillation that reduces test-time errors (i.e., discretization error in few-step sampling and exposure bias in causal generation). Extensive experiments across both bidirectional and causal architectures, at scales ranging from 1.3B to 14B parameters, demonstrate that AnyFlow achieves performance matches or surpasses consistency-based counterparts in the few-step regime, while scaling with sampling step budgets.

Text-to-image (T2I) generation has advanced rapidly, making reliable evaluation critical as performance differences between models narrow. Existing evaluation practices typically apply uniform annotation mechanisms, such as Likert-scale or binary question answering (BQA), across heterogeneous evaluation skills, despite fundamental differences in their nature. In this work, we revisit T2I evaluation through the lens of skill-aligned annotation, where annotation strategies reflect the underlying characteristics of each evaluation skill. We systematically compare skill-aligned annotation against uniform baselines and show that it produces more consistent evaluation signals, with higher inter-annotator agreement and improved stability across models. Finally, we present an automated pipeline that instantiates the proposed evaluation protocol, enabling scalable and fine-grained evaluation with spatially grounded feedback. Our work highlights that improving the foundations of image evaluation can increase reliability and efficiency without simply scaling annotation effort. We hope this motivates further research on refining evaluation protocols as a central component of reliable model assessment.

Diffusion-based models have shown strong performance in video super-resolution (VSR) and video frame interpolation (VFI). However, their role in the coupled space-time video super-resolution (STVSR) setting remains limited. Existing diffusion-based STVSR approaches suffer from two issues: (1) low inference efficiency and (2) insufficient utilization of spatiotemporal information. These limitations impede deployment. To address these issues, we introduce DiffST, an efficient spatiotemporal-aware video diffusion framework for real-world STVSR. To improve efficiency, we adapt a pre-trained diffusion model for one-step sampling and process the entire video directly rather than operating on individual frames. Furthermore, to enhance spatiotemporal information utilization, we introduce cross-frame context aggregation (CFCA) and video representation guidance (VRG). The CFCA module aggregates information across multiple keyframes to produce intermediate frames. The VRG module extracts video-level global features to guide the diffusion process. Extensive experiments show that DiffST obtains leading results on real-world STVSR tasks. It also maintains high inference efficiency, running about 17$\times$ faster than previous diffusion-based STVSR methods. Code is available at: https://github.com/zhengchen1999/DiffST.

Autoregressive video generation enables streaming and open-ended long video synthesis, but still suffers from long-term degradation caused by accumulated errors. Existing KVCache strategies usually apply unified historical-frame retention, implicitly assuming homogeneous historical dependencies across attention heads. We revisit historical-frame attention and reveal three distinct head types: Anchor Heads require broad long-range context, Wave Heads exhibit periodic temporal dependencies, and Veil Heads focus on initial and adjacent frames. Based on this finding, we propose Pyramid Forcing, a head-aware pyramidal KVCache framework that identifies head types offline, assigns behavior-specific cache policies, and supports heterogeneous cache lengths via efficient ragged-cache attention. Experiments on Self Forcing and Causal Forcing show that Pyramid Forcing consistently improves long-horizon generation quality on VBench-Long, increasing the 60-second Self Forcing score from 77.87 to 81.21 while enhancing motion dynamics, visual fidelity, and semantic consistency. Project: https://if-lab-pku.github.io/Pyramid-Forcing/.

Camera-conditioned video generation requires positional encoding that remains reliable under changes in camera motion, lens configuration, and scene structure. However, existing attention-level camera encodings either provide ray-only camera signals or rely on pinhole camera geometry, limiting their applicability to general camera control under the Unified Camera Model, including wide-angle and fisheye lenses. To address this limitation, we propose Curved Ray Expectation Positional Encoding (CRePE). CRePE represents each image token as a depth-aware positional distribution along its source ray, providing a Unified Camera Model-compatible positional encoding that captures the projected-path geometry induced by wide-angle and fisheye cameras. CRePE is implemented through a Geometric Attention Adapter added to frozen video DiTs, injecting token-wise scene-distance information into selected attention layers and stabilizing it with pseudo supervision from a monocular geometry foundation model. This design leads to more stable camera control and improves several geometry-aware and perceptual-quality metrics, while remaining competitive on video-quality metrics. Controlled positional-encoding ablations show a better overall average rank than a RayRoPE-style endpoint PE baseline, demonstrating the effectiveness of UCM-aware projected-path integration across diverse camera models. Furthermore, by extending the same positional-encoding pathway to external geometry control through Radial MixForcing, CRePE supports external radial-map control for scene-geometry-conditioned generation and source-video motion transfer beyond camera control.

Autoregressive video generation aims at real-time, open-ended synthesis. Yet, cinematic storytelling is not merely the endless extension of a single scene; it requires progressing through evolving events, viewpoint shifts, and discrete shot boundaries. Existing autoregressive models often struggle in this setting. Trained primarily for short-horizon continuation, they treat long sequences as extended single shots, inevitably suffering from motion stagnation and semantic drift during long rollouts. To bridge this gap, we introduce CausalCine, an interactive autoregressive framework that transforms multi-shot video generation into an online directing process. CausalCine generates causally across shot changes, accepts dynamic prompts on the fly, and reuses context without regenerating previous shots. To achieve this, we first train a causal base model on native multi-shot sequences to learn complex shot transitions prior to acceleration. We then propose Content-Aware Memory Routing (CAMR), which dynamically retrieves historical KV entries according to attention-based relevance scores rather than temporal proximity, preserving cross-shot coherence under bounded active memory. Finally, we distill the causal base model into a few-step generator for real-time interactive generation. Extensive experiments demonstrate that CausalCine significantly outperforms autoregressive baselines and approaches the capability of bidirectional models while unlocking the streaming interactivity of causal generation. Demo available at https://yihao-meng.github.io/CausalCine/

Dense 3D tracking from monocular video is fundamental to dynamic scene understanding. While recent 3D foundation models provide reliable per-frame geometry, recovering object motion in this geometry remains challenging and benefits from strong motion priors learned from real-world videos. Existing 3D trackers either follow iterative paradigms trained from scratch on synthetic data or fine-tune 3D reconstruction models learned from static multi-view images, both lacking real-world motion priors. Pre-trained video diffusion transformers (video DiTs) offer rich spatio-temporal priors from internet-scale videos, making them a promising foundation for 3D tracking. However, their frame-anchored formulation, which generates each frame's content, is fundamentally mismatched with reference-anchored dense 3D tracking, which must follow the same physical points from a reference frame across time. We present TrackCraft3R, the first method to repurpose a video DiT as a feed-forward dense 3D tracker. Given a monocular video and its frame-anchored reconstruction pointmap, TrackCraft3R predicts a reference-anchored tracking pointmap that follows every pixel of the first frame across time in a single forward pass, along with its visibility. We achieve this through two designs: (i) a dual-latent representation that uses per-frame geometry latents and reference-anchored track latents as dense queries, and (ii) temporal RoPE alignment, which specifies the target timestamp of each track latent. Together, these designs convert the per-frame generative paradigm of video DiTs into a reference-anchored tracking formulation with LoRA fine-tuning. TrackCraft3R achieves state-of-the-art performance on standard sparse and dense 3D tracking benchmarks, while running 1.3x faster and using 4.6x less peak memory than the strongest prior method. We further demonstrate robustness to large motions and long videos.

Recent advances in joint audio-video generation have been remarkable, yet real-world applications demand strong per-modality fidelity, cross-modal alignment, and fine-grained synchronization. Reinforcement Learning (RL) offers a promising paradigm, but its extension to multi-objective and multi-modal joint audio-video generation remains unexplored. Notably, our in-depth analysis first reveals that the primary obstacles to applying RL in this stem from: (i) multi-objective advantages inconsistency, where the advantages of multimodal outputs are not always consistent within a group; (ii) multi-modal gradients imbalance, where video-branch gradients leak into shallow audio layers responsible for intra-modal generation; (iii) uniform credit assignment, where fine-grained cross-modal alignment regions fail to get efficient exploration. These shortcomings suggest that vanilla RL fine-tuning strategy with a single global advantage often leads to suboptimal results. To address these challenges, we propose OmniNFT, a novel modality-aware online diffusion RL framework with three key innovations: (1) Modality-wise advantage routing, which routes independent per-reward advantages to their respective modality generation branches. (2) Layer-wise gradient surgery, which selectively detaches video-branch gradients on shallow audio layers while retaining those for cross-modal interaction layers. (3) Region-wise loss reweighting, which modulates policy optimization toward critical regions related to audio-video synchronization and fine-grained alignment. Extensive experiments on JavisBench and VBench with the LTX-2 backbone demonstrate that OmniNFT achieves comprehensive improvements in audio and video perceptual quality, cross-modal alignment, and audio-video synchronization.

Generative novel view synthesis faces a fundamental dilemma: geometric priors provide spatial alignment but become sparse and inaccurate under view changes, while appearance priors offer visual fidelity but lack geometric correspondence. Existing methods either propagate geometric errors throughout generation or suffer from signal conflicts when fusing both statically. We introduce MoCam, which employs structured denoising dynamics to orchestrate a coordinated progression from geometry to appearance within the diffusion process. MoCam first leverages geometric priors in early stages to anchor coarse structures and tolerate their incompleteness, then switches to appearance priors in later stages to actively correct geometric errors and refine details. This design naturally unifies static and dynamic view synthesis by temporally decoupling geometric alignment and appearance refinement within the diffusion process. Experiments demonstrate that MoCam significantly outperforms prior methods, particularly when point clouds contain severe holes or distortions, achieving robust geometry-appearance disentanglement.

Differentiable vector graphics have enabled powerful gradient-based optimization of vector primitives directly from raster images. However, existing frameworks formulate this as a flat optimization problem, forcing hundreds to thousands of randomly initialized curves to blindly compete for pixel-level error reduction. This disordered optimization leads to topology collapse, where macroscopic structures are distorted by internal high-frequency noise, resulting in a redundant and uneditable "polygon soup" that limits practical editability. To address this limitation, we propose Vector Scaffolding, a novel hierarchical optimization framework that shifts from flat pixel-matching to structured topological construction tailored for vector graphics. By identifying a key cause of topology collapse as the mathematical imbalance between area and boundary gradients, we introduce Interior Gradient Aggregation to stabilize the learning dynamics of multi-scale curve mixtures. Upon this stabilized landscape, we employ Progressive Stratification and Rapid Inflation Scheduling to progressively densify vector primitives with extremely high learning rates ($\times 50$). Experiments demonstrate that our approach accelerates optimization by $2.5\times$ while simultaneously improving PSNR by up to 1.4 dB over the previous state of the art.

Training-free camera control for pretrained flow-matching video generators is a partial-observation inverse problem: a depth-warped guidance video supplies noisy evidence on a subset of latent sites, which the sampler must reconcile with the pretrained prior. Existing methods struggle to balance the trade-off between trajectory adherence and visual quality and the heuristic guidance-strength tuning lacks robustness. We propose \textbf{$h$-control}, which resolves this dilemma through a structural change to the sampler: each outer hard-replacement guidance step is augmented with an inner-loop \emph{block-conditional pseudo-Gibbs refinement} on the unobserved complement at the same noise level, with provable convergence to the partial-observation conditional data law. To accelerate convergence on high-dimensional video latents, we exploit their conditional locality, partitioning the unobserved complement into 3D patches, each tracked by a custom mixing indicator that adaptively freezes converged patches. On RealEstate10K and DAVIS, \textbf{$h$-control} attains the best FVD against all seven training-free and training-based competitors, outperforming every training-free baseline on every reported metric.

While the overall inference latency of Video Diffusion Transformers (DiTs) can be substantially reduced through model distillation, per-step inference latency remains a critical bottleneck. Existing acceleration paradigms primarily exploit redundancy across the denoising trajectory; however, we identify a limitation where these step-wise strategies encounter diminishing returns in few-step regimes. In such scenarios, the scarcity of temporal states prevents effective feature reuse or predictive modeling, creating a formidable barrier to further acceleration. To overcome this, we propose Frame Interleaved Sparsity DiT (FIS-DiT), a training-free and operator-agnostic framework that shifts the optimization focus from the temporal trajectory to the latent frame dimension. Our approach is motivated by an intrinsic duality within this dimension: the existence of frame-wise sparsity that permits reduced computation, coupled with a structural consistency where each frame position remains equally vital to the global spatiotemporal context. Leveraging this insight, we implement Frame Interleaved Sparsity (FIS) as an execution strategy that manipulates frame subsets across the model hierarchy, refreshing all latent positions without requiring full-scale block computation. Empirical evaluations on Wan 2.2 and HunyuanVideo 1.5 demonstrate that FIS-DiT consistently achieves 2.11--2.41$\times$ speedup with negligible degradation across VBench-Q and CLIP metrics, providing a scalable and robust pathway toward real-time high-definition video generation.

Recent generative methods for single-shot high dynamic range (HDR) image reconstruction show promising results, but often struggle with preserving fidelity to the input image. They require separate models to handle highlights and shadows, or sacrifice interpretability by directly predicting the final HDR image. We address these limitations by re-casting single-shot HDR reconstruction as conditional video generation and fusing the generated frames into an HDR image. We finetune a video diffusion model to generate an exposure bracket, conditioned on a low dynamic range (LDR) input. We fuse this image bracket using per-pixel weights predicted by a light-weight UNet. This formulation is simple, interpretable, and effective. Rather than directly hallucinating an HDR image, it explicitly reconstructs the intermediate exposure stack and fuses it into the final output. Our method eliminates the need for separate models across exposure regimes and produces HDR reconstructions with high input fidelity. On quantitative benchmarks, we outperform state-of-the-art generative baselines with comparable model capacity on several reconstruction metrics. Human evaluators further prefer our results in 72% of pairwise comparisons against existing methods. Finally, we show that this input-conditioned sequence generation and fusion framework extends beyond HDR to other image reconstruction tasks, such as all-in-focus image recovery from a single defocus-blurred input.

Today's driving world models can generate remarkably realistic dash-cam videos, yet no single model excels universally. Some generate photorealistic textures but violate basic physics; others maintain geometric consistency but fail when subjected to closed-loop planning. This disconnect exposes a critical gap: the field evaluates how real generated worlds appear, but rarely whether they behave realistically. We introduce WorldLens, a unified benchmark that measures world-model fidelity across the full spectrum, from pixel quality and 4D geometry to closed-loop driving and human perceptual alignment, through five complementary aspects and 24 standardized dimensions. Our evaluation of six representative models reveals that no existing approach dominates across all axes: texture-rich models violate geometry, geometry-aware models lack behavioral fidelity, and even the strongest performers achieve only 2-3 out of 10 on human realism ratings. To bridge algorithmic metrics with human perception, we further contribute WorldLens-26K, a 26,808-entry human-annotated preference dataset pairing numerical scores with textual rationales, and WorldLens-Agent, a vision-language evaluator distilled from these judgments that enables scalable, explainable auto-assessment. Together, the benchmark, dataset, and agent form a unified ecosystem for assessing generated worlds not merely by visual appeal, but by physical and behavioral fidelity.

Generative world models are increasingly used for video generation, where learned simulators are expected to capture the physical rules that govern real-world dynamics. However, evaluating whether generated videos actually follow these rules remains challenging. Existing physics-focused video benchmarks have made important progress, but they still face three key challenges, including the coarse evaluation frameworks that hide law-specific failures, response biases and fatigue that undermine the validity of annotation judgments, and automated evaluators that are insufficiently physics-aware or difficult to audit. To address those challenges, we introduce PhyGround, a criteria-grounded benchmark for evaluating physical reasoning in video generation. The benchmark contains 250 curated prompts, each augmented with an expected physical outcome, and a taxonomy of 13 physical laws across solid-body mechanics, fluid dynamics, and optics. Each law is operationalized through observable sub-questions to enable per-law diagnostics. We evaluate eight modern video generation models through a large-scale, quality-controlled human study, grounded on social science lab experiment design. A total of 459 annotators provided 5,796 complete annotations and over 37.4K fine-grained labels; after quality control, the retained annotations exhibited high split-half model-ranking correlations (Spearman's rho > 0.90). To support reproducible automated evaluation, we release PhyJudge-9B, an open physics-specialized VLM judge. PhyJudge-9B achieves substantially lower aggregate relative bias than Gemini-3.1-Pro (3.3% vs. 16.6%). We release prompts, human annotations, model checkpoints, and evaluation code on the project page https://phyground.github.io/.

Commercial video generation systems such as Seedance2.0 and Veo3.1 have rapidly improved, strengthening the view that video generators may be evolving into "world simulators." Yet the community still lacks a benchmark that directly tests whether a model can reason about how an observed world should evolve over time. We introduce WorldReasonBench, which reframes video generation evaluation as world-state prediction: given an initial state and an action, can a model generate a future video whose state evolution remains physically, socially, logically, and informationally consistent? WorldReasonBench contains 436 curated test cases with structured ground-truth QA annotations spanning four reasoning dimensions and 22 subcategories. We evaluate generated videos with a human-aligned two-part methodology: Process-aware Reasoning Verification uses structured QA and reasoning-phase diagnostics to detect temporal and causal failures, while Multi-dimensional Quality Assessment scores reasoning quality, temporal consistency, and visual aesthetics for ranking and reward modeling. We further introduce WorldRewardBench, a preference benchmark with approximately 6K expert-annotated pairs over 1.4K videos, supporting pair-wise and point-wise reward-model evaluation. Across modern video generators, our results expose a persistent gap between visual plausibility and world reasoning: videos can look convincing while failing dynamics, causality, or information preservation. We will release our benchmarks and evaluation toolkit to support community research on genuinely world-aware video generation at https://github.com/UniX-AI-Lab/WorldReasonBench/.

Video generation has advanced rapidly, producing photorealistic videos from text or image prompts. Meanwhile, film production and social robotics increasingly demand multi-person videos with rich social interactions, including conversations, gestures, and coordinated actions. However, existing models offer no explicit control over interactions, such as who performs which action, when it occurs, and toward whom it is directed. This often results in wrong person performing unintended actions (actor-action mismatch), disordered social dynamics, and wrong action targets. To address these challenges, we present SocialDirector, a training-free interaction controller that enhances the generation model by modulating cross-attention maps. SocialDirector contains two modules: Social Actor Masking and Directional Reweighting. Social Actor Masking constrains each person's visual tokens to attend only to their own textual descriptions via a spatiotemporal mask, avoiding actor-action mismatch and disordered social dynamics. Directional Reweighting amplifies attention to directional words (e.g., "leftward", "right"), leading each action towards its intended target. To evaluate generated social interactions, we annotate existing datasets with interaction descriptions and build a fully automated evaluation pipeline powered by open-source VLMs. Experiments on different video generation models show that SocialDirector significantly improves interaction fidelity and approaches the upper bound set by real videos.

Autoregressive (AR) video diffusion models adopt a streaming generation framework, enabling long-horizon video generation with real-time responsiveness, as exemplified by the Self Forcing training paradigm. However, existing AR video diffusion models still suffer from significant attention complexity and severe memory overhead due to the redundant key-value (KV) caches across historical frames, which limits scalability. In this paper, we tackle this challenge by introducing KV cache compression into autoregressive video diffusion. We observe that attention heads in mainstream AR diffusion models exhibit markedly distinct attention patterns and functional roles that remain stable across samples and denoising steps. Building on our empirical study of head-wise functional specialization, we divide the attention heads into two categories: static heads, which focus on transitions across autoregressive chunks and intra-frame fidelity, and dynamic heads, which govern inter-frame motion and consistency. We then propose Forcing-KV, a hybrid KV cache compression strategy that performs structured static pruning for static heads and dynamic pruning based on segment-wise similarity for dynamic heads. While maintaining output quality, our method achieves a generation speed of over 29 frames per second on a single NVIDIA H200 GPU along with 30% cache memory reduction, delivering up to 1.35x and 1.50x speedups on LongLive and Self Forcing at 480P resolution, and further scaling to 2.82x speedup at 1080P resolution. Code and demo videos are provided at https://zju-jiyicheng.github.io/Forcing-KV-Page.

Teleoperation systems are fundamentally limited by communication latency, which degrades situational awareness and control performance. Predictive display aims to mitigate this limitation by presenting an estimate of the current visual state rather than delayed observations. While recent advances in generative video models enable high-quality video synthesis, their suitability for latency-sensitive predictive display remains unclear. This paper presents a zero-shot benchmark of off-the-shelf generative video models for short-horizon predictive display, without task-specific fine-tuning. We formulate the problem as rollout-based future frame prediction and develop a unified benchmarking pipeline using simulated driving data from the CARLA simulator. Five publicly released video models spanning transformer-based and diffusion-based families are evaluated across two resolutions and two conditioning regimes (multi-frame and single-frame). Performance is assessed using prediction accuracy (mean absolute difference), per-rollout latency, peak GPU memory usage, and temporal error evolution across the prediction horizon. On this zero-shot benchmark, no tested model simultaneously achieves low rollout error, non-divergent per-step error behavior, and real-time inference at the source frame rate. Increasing model scale or resolution yields limited and, in some cases, inverted improvements. These findings highlight a gap between general-purpose generative video synthesis and the requirements of predictive display in teleoperation, suggesting that practical deployment will require either explicit short-horizon temporal supervision, in-domain adaptation, or aggressive inference optimization rather than direct application of off-the-shelf models. Code, configurations, and qualitative results are released on the project page: https://bimilab.github.io/paper-GenPD

Hair is a rich medium of visual and cultural expression, yet its digital modeling remains challenging due to the duality of fluidity and structure. Many existing generative approaches rely primarily on continuous diffusion fields, which entangle global topology with local texture and obscure the semantic and structural organization of hairstyles. To address this, we propose HairGPT, a strand-centric framework that treats strands as generative primitives and formulates realistic 3D hairstyle synthesis as a dual-decoupled autoregressive sequence modeling problem. Our method applies spatial decoupling across semantic scalp regions and structural decoupling along a hierarchical strand representation, progressing from global layout to fine-grained style. We further introduce a geometric tokenizer and region-aware semantic annotations to guide strand-level generation, enabling compositional editing, synthesis of rare and complex hairstyles, and adaptation to stylized domains. By aligning generative modeling with the workflow of digital grooming, HairGPT turns hair generation from opaque texture synthesis into a structured and semantically controllable authoring process, supporting robust semantic conditioning and high-fidelity results across realistic and stylized domains. Project Page: https://haiminluo.github.io/hairgpt/

Action-conditioned surgical video generation is a critical yet highly challenging problem for robotic surgery. The core difficulty is that low-dimensional control vectors must precisely govern complex image-space evolution. In this work, we propose a kinematic-to-visual lifting paradigm that converts articulated kinematics into a unified set of five image-aligned control modalities. Building on this representation, we introduce a hierarchically routed visual control framework that selectively activates the most relevant control modalities and motion scales. Instead of uniformly applying all control signals, our model performs hierarchical routing to dynamically allocate conditioning capacity. We further design kinematic-prior-guided routing loss functions to ensure physically meaningful, temporally stable, and efficient expert utilization. To improve efficiency, we propose a budgeted training and inference scheme that leverages routing-induced sparsity. By selectively discarding low-significance control pathways during training and execution, our approach enables adaptive computation that is complementary to standard distillation. We additionally construct a new benchmark with curated articulated annotations, obtained through human-in-the-loop semantic labeling and differentiable pose tracking, providing realistic supervision for action-conditioned surgical video generation. Extensive experiments demonstrate that our method consistently improves action faithfulness, visual fidelity, and cross-domain generalization over diverse baselines. Moreover, our efficient variant achieves substantial reductions in latency while maintaining strong control accuracy.

Generating human motion that satisfies customized zero-shot goal functions, enabling applications such as controllable character animation and behavior synthesis for virtual agents, is a critical capability. While current approaches handle many unseen constraints, they fail on tasks with very challenging spatiotemporal restrictions, such as severe spatial obstacles or specified numbers of walking steps. To equip motion generators for these highly constrained tasks, we present a retrieval-guided method built on the training-free diffusion noise optimization framework. The key idea is to search within large motion datasets for guidance that can potentially satisfy difficult constraints. We introduce relational task parsing to group target constraints and identify the difficult ones to be handled by retrieved reference. A better initialization for diffusion noise is then obtained via a reward-guided mask that combines random noise with retrieved noise. By optimizing diffusion noise from this improved initialization, we successfully solve highly constrained generation tasks. By leveraging LLM for relational task parsing, the whole framework is further enabled to automatically reason for what to retrieve, improving the intelligence of moving agents under a training-free optimization scheme.

Talking-head generation requires joint modeling of identity, head pose, facial expression, and mouth dynamics. Existing methods typically address only a subset of these factors, and rely on fixed-weight or heuristic fusion when multiple conditions are involved. We present MoCoTalk, a multi-conditional video diffusion framework that unifies four complementary control signals: a reference image, facial keypoints, 3DMM-rendered shading meshes, and the corresponding speech audio. To resolve destructive interference among heterogeneous conditions, we introduce an Adaptive Multi-Condition Router that computes channel-wise, timestep-aware gating over the four condition streams, allowing the fusion strategy to vary with both feature subspace and noise level. To better capture speech-related facial dynamics, we design a Mouth-Augmented Shading Mesh, a 3DMM-based representation that decouples head motion, mouth motion, expression, and lighting. This design provides a temporally consistent geometric prior and allows flexible recombination of these attributes at inference. We further introduce a lip consistency loss to tighten audio-visual alignment. Extensive experiments show that MoCoTalk achieves state-of-the-art performance on the majority of structural, motion, and perceptual metrics, while offering attribute-level controllability that single-condition methods do not provide.

Recent video diffusion models (VDMs) synthesize visually convincing clips, yet still drop entities, mis-bind attributes, and weaken the interactions specified in the prompt. Representation-alignment objectives such as VideoREPA and MoAlign improve fine-grained text following by distilling spatio-temporal token relations from a frozen visual foundation model, but their pairwise supervision budget is allocated by visual or motion cues rather than by how relevant each pair is to the prompt. We present SARA, Semantically Adaptive Relational Alignment, which keeps token-relation distillation (TRD) on a frozen VFM target and adds a text-conditioned saliency that decides which token pairs carry supervision. A lightweight Stage1 aligner is trained with per-entity SAM3.1 mask supervision and an InfoNCE regulariser, and its continuous saliency is fused into TRD through a pair-routing operator that assigns each token pair a weight whenever either of its two endpoints is salient, thereby routing supervision toward subject-subject and subject-background pairs and away from background-background ones. In the Wan2.2 continual-training setting, SARA improves both text alignment and motion quality over SFT, VideoREPA, and MoAlign on a 13-dimension VLM rubric, on the public VBench benchmarks, and in a blind user study. Project page: https://saradit.github.io/.

Efficiently aligning large-scale video diffusion models with human intent requires a scalable and trajectory-aware pathway that bridges the inherent discrepancy between training noise distributions and practical inference trajectories. While existing paradigms such as Direct Preference Optimization (DPO) and Group Relative Policy Optimization (GRPO) attempt to address this, they are often hindered by either reliance on bias-prone, complex reward models or suboptimal timestep sampling. In this paper, we propose Diffusion-APO (Aligned Preference Optimization), a trajectory-aware algorithm that resolves this misalignment by synchronizing training noise with inference-time denoising paths to maximize gradient signal efficacy. To translate this algorithmic innovation into a practical solution, we introduce a unified and modular RLHF framework that integrates online ranking, half-online anchoring, offline refinement, and distillation-aware drift correction. This framework enables flexible, multi-stage preference alignment across diverse data and computational constraints without relying on scalar-reward-based policy gradients. Through extensive experiments, we demonstrate that Diffusion-APO consistently outperforms standard baselines in visual quality and instruction following, while effectively preserving generative fidelity during model acceleration, providing a robust, end-to-end pathway for scalable video diffusion alignment.

Recent advances in generative AI have democratized video creation at scale. AI-generated videos, including partially manipulated clips across visual and audio channels, pose escalating risks of semantic distortion and misuse, which motivates the need for reliable detection tools. Most existing AI-generated video detectors remain limited by single- or partial-modality of data modeling and the lack of fine-grained temporal forgery localization. To address these challenges, our primary novelty introduces a core architecture that jointly integrates an LMM semantic branch with a spatio-temporal (ST) visual branch and a multi-scale partial-spoof (PS) audio branch. This multi-modal approach enables simultaneous detection and fine-grained temporal localization of partially manipulated AI-generated video forgeries. Extensive experiments show that this approach outperforms existing state-of-the-art methods.

Synthesizing consistent and coherent long video remains a fundamental challenge. Existing methods suffer from semantic drift and narrative collapse over long horizons. We present A$^2$RD, an Agentic Auto-Regressive Diffusion architecture that decouples creative synthesis from consistency enforcement. A$^2$RD formulates long video synthesis as a closed-loop process that synthesizes and self-improves video segment-by-segment through a Retrieve--Synthesize--Refine--Update cycle. It comprises three core components: (i) Multimodal Video Memory that tracks video progression across modalities; (ii) Adaptive Segment Generation that switches among generation modes for natural progression and visual consistency; and (iii) Hierarchical Test-Time Self-Improvement that self-improves each segment at frame and video levels to prevent error propagation. We further introduce LVBench-C, a challenging benchmark with non-linear entity and environment transitions to stress-test long-horizon consistency. Across public and LVBench-C benchmarks spanning one- to ten-minute videos, A$^2$RD outperforms state-of-the-art baselines by up to 30% in consistency and 20% in narrative coherence. Human evaluations corroborate these gains while also highlighting notable improvements in motion and transition smoothness.

Diffusion Transformers (DiTs) have achieved state-of-the-art video generation quality, but they incur immense computational cost because standard inference applies the same number of denoising steps uniformly to every token in the sequence. It is well known that human vision ignores vast amounts of redundant motion. Why, then, do our densest models treat every spatiotemporal token with equal priority? In this paper, we introduce Heterogeneous Step Allocation (HSA), a training-free inference algorithm that assigns varying step budgets to different spatiotemporal tokens based on their velocity dynamics. To resolve the resulting sequence-length mismatch without sacrificing global context, HSA introduces a KV-cache synchronization mechanism that allows active tokens to attend to the full sequence while entirely bypassing inactive tokens. Furthermore, we derive a cached Euler update that advances the latent states of skipped tokens in a single operation without additional model evaluations. We evaluate HSA on the Wan-2 and LTX-2 models for both text-to-video (T2V) and image-to-video (I2V) generation. Our results demonstrate that HSA significantly outperforms previous state-of-the-art caching methods and the vanilla Flow Matching baseline, especially at aggressive acceleration regimes (e.g., 50% and 25% runtimes). Crucially, HSA achieves a superior quality-runtime Pareto frontier without the need for expensive offline profiling, robustly preserving structural integrity and generation quality even under tight computational budgets. Project page: https://ernestchu.github.io/hsa

For artistic applications, video generation requires fine-grained control over both performance and cinematography, i.e., the actor's motion and the camera trajectory. We present ActCam, a zero-shot method for video generation that jointly transfers character motion from a driving video into a new scene and enables per-frame control of intrinsic and extrinsic camera parameters. ActCam builds on any pretrained image-to-video diffusion model that accepts conditioning in terms of scene depth and character pose. Given a source video with a moving character and a target camera motion, ActCam generates pose and depth conditions that remain geometrically consistent across frames. We then run a single sampling process with a two-phase conditioning schedule: early denoising steps condition on both pose and sparse depth to enforce scene structure, after which depth is dropped and pose-only guidance refines high-frequency details without over-constraining the generation. We evaluate ActCam on multiple benchmarks spanning diverse character motions and challenging viewpoint changes. We find that, compared to pose-only control and other pose and camera methods, ActCam improves camera adherence and motion fidelity, and is preferred in human evaluations, especially under large viewpoint changes. Our results highlight that careful camera-consistent conditioning and staged guidance can enable strong joint camera and motion control without training. Project page: https://elkhomar.github.io/actcam/.

Recent advances have shown that large-scale video diffusion models can be repurposed as neural renderers by first decomposing videos into intrinsic scene representations and then performing forward rendering under novel illumination. While promising, this paradigm fundamentally relies on accurate intrinsic decomposition, which remains highly unreliable for real-world videos and often leads to distorted appearances, broken materials, and accumulated temporal artifacts during relighting. In this work, we present Relit-LiVE, a novel video relighting framework that produces physically consistent, temporally stable results without requiring prior knowledge of camera pose. Our key insight is to explicitly introduce raw reference images into the rendering process, enabling the model to recover critical scene cues that are inevitably lost or corrupted in intrinsic representations. Furthermore, we propose a novel environment video prediction formulation that simultaneously generates relit videos and per-frame environment maps aligned with each camera viewpoint in a single diffusion process. This joint prediction enforces strong geometric-illumination alignment and naturally supports dynamic lighting and camera motion, significantly improving physical consistency in video relighting while easing the requirement of known per-frame camera pose. Extensive experiments demonstrate that Relit-LiVE consistently outperforms state-of-the-art video relighting and neural rendering methods across synthetic and real-world benchmarks. Beyond relighting, our framework naturally supports a wide range of downstream applications, including scene-level rendering, material editing, object insertion, and streaming video relighting. The Project is available at https://github.com/zhuxing0/Relit-LiVE.

Video diffusion models perform well in short-video synthesis, but their training-free extension to long videos often suffers from content drift, temporal inconsistency, and over-smoothed dynamics. Existing methods improve temporal consistency by combining a global branch with a local branch, but they often further decompose appearance consistency and temporal dynamics within each branch using predefined criteria. This assignment is unreliable when appearance and action progression are tightly coupled, such as in camera motion and sequential motion. We analyze the video temporal extension issue from a singular-spectrum perspective and show that enlarged self-attention windows induce spectral concentration: spectral energy becomes dominated by a few low-rank singular directions, preserving coarse structure but suppressing high-rank spatial details and motion-rich temporal variations. To mitigate this problem, we propose FreeSpec, a training-free spectral reconstruction framework for long-video generation. FreeSpec decomposes global and local features with singular value decomposition, and uses the global branch as low-rank spectral guidance and the local branch as a high-rank reconstruction basis. This spectrum-level fusion avoids the rigid feature partitioning of previous decomposition rules, preserving long-range consistency while better retaining spatial details and temporal dynamics. Experiments on Wan2.1 and LTX-Video demonstrate that FreeSpec improves long-video generation, especially for temporal dynamics, while maintaining strong visual quality and temporal consistency. Project demo: https://fdchen24.github.io/FreeSpec-Website/.

Recent advancements in image animation have utilized diffusion models to breathe life into static images. However, existing controllable frameworks typically rely on Lagrangian motion guidance, where optical flow is estimated relative to the initial frame. This paper revisits the same optical-flow primitive through a more local supervision design: we use adjacent-frame Eulerian motion fields to guide generation, where the motion signal always describes a short temporal hop. This shift enables parallelized training and provides bounded-error supervision throughout the generation process. To mitigate the drift artifacts common in adjacent frame generation, we introduce a Bidirectional Geometric Consistency mechanism, which computes a forward-backward cycle check to mathematically identify and mask occluded regions, preventing the model from learning incorrect warping objectives. Extensive experiments demonstrate that our approach accelerates training, preserves temporal coherence, and reduces dynamic artifacts compared to reference-based baselines. The code, model, and data have been made available at https://nguyentthong.github.io/eulerian/

Existing text-to-image (T2I) benchmarks largely rely on fixed prompt sets, leaving them vulnerable to overfitting and benchmark contamination once publicly released and repeatedly reused. In this work, we propose DynT2I-Eval, a fully automated dynamic evaluation framework for T2I models. It constructs a structured visual semantic space from long-form descriptions, decomposing prompts into controllable dimensions (e.g., subject, logical constraint, environment, and composition). This enables the continuous generation of fresh prompts via task-specific spaces and difficulty-aware sampling. DynT2I-Eval evaluates model performance across text alignment, perceptual quality, and aesthetics. Heterogeneous outputs are unified into prompt-conditioned pairwise comparisons, allowing a dynamic scheduler, micro-batch aggregation, and weighted Bayesian updates to maintain a stable online leaderboard despite changing prompt distributions and model injection. Experiments with independently sampled prompt streams demonstrate that continually refreshed prompts provide a robust evaluation protocol, reducing the impact of prompt-set-specific tuning. Simulations and ablations further confirm that the proposed ranking framework achieves a strong balance among cold-start convergence, late-entry discovery, and long-run ranking fidelity.

Camera-controlled video-to-video (V2V) generation enables dynamic viewpoint synthesis from monocular footage, holding immense potential for interactive filmmaking and live broadcasting. However, existing implicit synthesis methods fundamentally rely on non-causal, full-sequence processing and rigid prefix-style temporal concatenation. This architectural paradigm mandates bidirectional attention, resulting in prohibitive computational latency, quadratic complexity scaling, and inherent incompatibility with real-time streaming or variable-length inputs. To overcome these limitations, we introduce \texttt{RealCam}, a novel autoregressive framework for interactive, real-time camera-controlled V2V generation. We first design a high-fidelity teacher model grounded in a \textbf{Cross-frame In-context Learning} paradigm. By interleaving source and target frames into synchronized contextual pairs, our design inherently enables length-agnostic generalization and naturally facilitates causal adaptation, breaking the rigid prefix bottleneck. We then distill this teacher into a few-step causal student via Self-Forcing with Distribution Matching Distillation, enabling efficient, on-the-fly streaming synthesis. Furthermore, to mitigate severe loop inconsistency in closed-loop trajectories, we propose \textbf{Loop-Closed Data Augmentation (LoopAug)}, a novel paradigm that synthesizes globally consistent loop sequences from existing multiview datasets. Extensive experiments demonstrate that \texttt{RealCam} achieves state-of-the-art visual fidelity and temporal consistency while enabling truly interactive camera control with orders-of-magnitude faster inference than existing paradigms. Our project page is at https://xyc-fly.github.io/RealCam/.

Recent advances in generative video models are increasingly driven by post-training and test-time scaling, both of which critically depend on the quality of video reward models (RMs). An ideal reward model should predict accurate rewards that align with human preferences across diverse scenarios. However, existing paradigms face a fundamental dilemma: \textit{Discriminative RMs} regress rewards directly on features extracted by multimodal large language models (MLLMs) without explicit reasoning, making them prone to shortcut learning and heavily reliant on massive data scaling for generalization. In contrast, \textit{Generative RMs} with Chain-of-Thought (CoT) reasoning exhibit superior interpretability and generalization potential, as they leverage fine-grained semantic supervision to internalize the rationales behind human preferences. However, they suffer from inherent optimization bottlenecks due to the coupling of reasoning and scoring within a single autoregressive inference chain. To harness the generalization benefits of CoT reasoning while mitigating the training instability of coupled reasoning and scoring, we introduce DeScore, a training-efficient and generalizable video reward model. DeScore employs a decoupled ``think-then-score'' paradigm: an MLLM first generates an explicit CoT, followed by a dedicated discriminative scoring module consisting of a learnable query token and a regression head that predicts the final reward. DeScore is optimized via a two-stage framework: (1) a discriminative cold start incorporating a random mask mechanism to ensure robust scoring capabilities, and (2) a dual-objective reinforcement learning stage that independently refines CoT reasoning quality and calibrates the final reward, ensuring that higher-quality reasoning directly translates to superior model performance.

Modern AI-generated videos are photorealistic at the single-frame level, leaving inter-frame dynamics as the main remaining axis for detection. Existing detectors typically handle this temporal evidence in three ways: feeding the full frame sequence to a generic temporal backbone, reducing one dominant temporal cue to fixed video-level descriptors, or comparing temporal features to real-video statistics through a detection metric. These strategies degrade sharply under cross-generator evaluation, where artifact type and timescale vary across generators. On caption-paired benchmark, GenVidBench, we identify two signatures that prior detectors do not jointly exploit: AI-generated videos exhibit smoother frame-to-frame temporal residuals at the pixel level, and more compact trajectories in the semantic feature space, indicating a temporal smoothness gap at both levels. We further observe that, when raw video is fed into a Spiking Neural Networks (SNNs), fake clips elicit firing predominantly at object and motion boundaries, unlike real clips, suggesting that the SNN responds to temporal artifacts localized at edges. These cues are sparse, asynchronous, and concentrated at moments of change, which makes SNNs a natural choice for this task: their event-driven, sparsely-activated dynamics align with the structure of the residual signal in a way that dense ANN backbones do not. Building on this observation, we propose MAST, a detector that processes multi-channel temporal residuals with a spike-driven temporal branch alongside a frozen semantic encoder for cross-generator generalization. On the GenVideo benchmark, MAST achieves 93.14% mean accuracy across 10 unseen generators under strict cross-generator evaluation, matching or surpassing the strongest ANN-based detectors and demonstrating the practical applicability of SNNs to AI-generated video detection.

This paper reports on the LoViF 2026 PhyScore challenge, a competition on holistic quality assessment of world-model-generated videos across both 2D and 4D generation settings. The challenge is motivated by a central gap in current evaluation practice: perceptual quality alone is insufficient to judge whether generated dynamics are physically plausible, temporally coherent, and consistent with input conditions. Participants are required to build a metric that jointly predicts four dimensions, i.e., Video Quality, Physical Realism, Condition-Video Alignment, and Temporal Consistency. Depart from that, participants also need to localize physical anomaly timestamps for fine-grained diagnosis. The benchmark dataset contains 1,554 videos generated by seven representative world generative models, organized into three tracks (text-2D, image-to-4D, and video-to-4D) and spanning 26 categories. These categories explicitly cover physics-relevant scenarios, including dynamics, optics, and thermodynamics, together with diverse real-world and creative content. To ensure label reliability, scores and anomaly timestamps are produced through trained human annotation with an additional automated quality-control pass. Evaluation is based on both score prediction and anomaly localization, with a composite protocol that combines TimeStamp_IOU and SRCC/PLCC. This report summarizes the challenge design and provides method-level insights from submitted solutions.

Video sequence capturing through refractive dynamic media, such as a turbulent air or water surface, often suffer from severe geometric distortions and temporal instability. While recent advances address mild atmospheric turbulence, no existing benchmarks systematically evaluate restoration methods under strong and highly nonuniform refractive conditions. We present a comprehensive benchmark for geometric distortion removal in video, covering a range from turbulence-like mild warping to strong discontinuous refractive deformations. The benchmark includes both laboratory-captured real data and synthetic sequences generated for static scenes via physics-based light refraction modeling across four distortion levels and multiple surface wave types. We evaluate a spectrum of methods from simple baselines and classical registration algorithms to advanced learning-based approaches including DATUM and our proposed diffusion based V-cache for high and extreme distortions regimes. Evaluation uses both pixel-level (PSNR, SSIM), and perceptual (LPIPS, DINO, CLIP) metrics providing the first large scale analysis of geometric distortion removal. Our benchmark establishes a new foundation for developing and evaluating algorithms capable of reconstructing video from highly distorted optical environments. Our code and datasets are available at https://github.com/iafoss/refractive-mfir-benchmark.

While Test-Time Scaling (TTS) offers a promising direction to enhance video generation without the surging costs of training, current test-time video generation methods based on diffusion models suffer from exorbitant candidate exploration costs and lack temporal guidance. To address these structural bottlenecks, we propose shifting the focus to streaming video generation. We identify that its chunk-level synthesis and few denoising steps are intrinsically suited for TTS, significantly lowering computational overhead while enabling fine-grained temporal control. Driven by this insight, we introduced Stream-T1, a pioneering comprehensive TTS framework exclusively tailored for streaming video generation. Specifically, Stream-T1 is composed of three units: (1) Stream -Scaled Noise Propagation, which actively refines the initial latent noise of the generating chunk using historically proven, high-quality previous chunk noise, effectively establishes temporal dependency and utilizing the historical Gaussian prior to guide the current generation; (2) Stream -Scaled Reward Pruning, which comprehensively evaluates generated candidates to strike an optimal balance between local spatial aesthetics and global temporal coherence by integrating immediate short-term assessments with sliding-window-based long-term evaluations; (3) Stream-Scaled Memory Sinking, which dynamically routes the context evicted from KV-cache into distinct updating pathways guided by the reward feedback, ensuring that previously generated visual information effectively anchors and guides the subsequent video stream. Evaluated on both 5s and 30s comprehensive video benchmarks, Stream-T1 demonstrates profound superiority, significantly improving temporal consistency, motion smoothness, and frame-level visual quality.

Distillation-based acceleration has become foundational for making autoregressive streaming video diffusion models practical, with distribution matching distillation (DMD) as the de facto choice. Existing methods, however, train the student to match the teacher's output indiscriminately, treating every rollout, frame, and pixel as equally reliable supervision. We argue that this caps distilled quality, since it overlooks two complementary axes of variance in DMD supervision: Inter-Reliability across student rollouts whose supervision varies in reliability, and Intra-Perplexity across spatial regions and temporal frames that contribute unequally to where quality can still be improved. The objective thus conflates two questions under a uniform weight: whether to learn from each rollout, and where to concentrate optimization within it. To address this, we propose Stream-R1, a Reliability-Perplexity Aware Reward Distillation framework that adaptively reweights the distillation objective at both rollout and spatiotemporal-element levels through a single shared reward-guided mechanism. At the Inter-Reliability level, Stream-R1 rescales each rollout's loss by an exponential of a pretrained video reward score, so that rollouts with reliable supervision dominate optimization. At the Intra-Perplexity level, it back-propagates the same reward model to extract per-pixel gradient saliency, which is factored into spatial and temporal weights that concentrate optimization pressure on regions and frames where refinement yields the largest expected gain. An adaptive balancing mechanism prevents any single quality axis from dominating across visual quality, motion quality, and text alignment. Stream-R1 attains consistent improvements on all three dimensions over distillation baselines on standard streaming video generation benchmarks, without architectural modification or additional inference cost.

Video generation models internalize physical realism as their prior. Anime deliberately violates physics: smears, impact frames, chibi shifts; and its thousands of coexisting artistic conventions yield no single "physics of anime" a model can absorb. Physics-biased models therefore flatten the artistry that defines the medium or collapse under its stylistic variance. We present AniMatrix, a video generation model that targets artistic rather than physical correctness through a dual-channel conditioning mechanism and a three-step transition: redefine correctness, override the physics prior, and distinguish art from failure. First, a Production Knowledge System encodes anime as a structured taxonomy of controllable production variables (Style, Motion, Camera, VFX), and AniCaption infers these variables from pixels as directorial directives. A trainable tag encoder preserves the field-value structure of this taxonomy while a frozen T5 encoder handles free-form narrative; dual-path injection (cross-attention for fine-grained control, AdaLN modulation for global enforcement) ensures categorical directives are never diluted by open-ended text. Second, a style-motion-deformation curriculum transitions the model from near-physical motion to full anime expressiveness. Third, deformation-aware preference optimization with a domain-specific reward model separates intentional artistry from pathological collapse. On an anime-specific human evaluation with five production dimensions scored by professional animators, AniMatrix ranks first on four of five, with the largest gains over Seedance-Pro 1.0 on Prompt Understanding (+0.70, +22.4 percent) and Artistic Motion (+0.55, +16.9 percent). We are preparing accompanying resources for public release to support reproducibility and follow-up research.

Evaluating generative video models remains an open problem. Reference-based metrics such as Structural Similarity Index Measure (SSIM) and Peak Signal to Noise Ratio (PSNR) reward pixel fidelity over semantic correctness, while Frechet Video Distance (FVD) favors distributional textures over physical plausibility. Binary Visual Question Answering (VQA) based benchmarks like VBench~2.0 are prone to yes-bias and rely on low-resolution auditors that miss temporal failures. Moreover, their prompts target a single dimension at a time, multiplying the number of videos required while still not guaranteeing reliable results. WorldJen addresses these limitations directly. Binary VQA is replaced with Likert-scale questionnaires graded by a VLM that receives frames at native video resolution. Video generation costs are addressed by using adversarially curated prompts that are designed to exercise up to 16 quality dimensions simultaneously. The framework is built around two interlocking contributions. First, A blind human preference study is conducted, accumulating (2,696 pairwise annotations from 7 annotators with 100% pair coverage over 50 of the curated prompts $\times$ 6 state-of-the-art video models. A mean inter-annotator agreement of 66.9% is achieved and the study establishes a human ground-truth Bradley-Terry (BT) rating with a three-tier structure. Second, A VLM-as-a-judge evaluation engine using prompt-specific, dimension-specific Likert questionnaires (10 questions per dimension, 47,160 scored responses) judges the videos and reproduces the human-established three-tier BT rating structure independently. The VLM achieves a Spearman $\hatρ=1.000,~p=0.0014$ that is interpreted as tier agreement with the human results. Six focused ablation studies validate the robustness of the VLM evaluation framework. Project page: https://moonmath.ai/worldjen/

We present Mamoda2.5, a unified AR-Diffusion framework that seamlessly integrates multimodal understanding and generation within a single architecture. To efficiently enhance the model's generation capability, we equip the Diffusion Transformer backbone with a fine-grained Mixture-of-Experts (MoE) design (128 experts, Top-8 routing), yielding a 25B-parameter model that activates only 3B parameters, significantly reducing training costs while scaling up the model capacity. Mamoda2.5 achieves top-tier generation performance on VBench 2.0 and sets a new record in video editing quality, surpassing evaluated open-source models and matching the performance of current top-tier proprietary models, including the Kling O1 on OpenVE-Bench. Furthermore, we introduce a joint few-step distillation and reinforcement learning framework that compresses the 30-step editing model into a 4-step model and greatly accelerates model inference. Compared to open-source baselines, Mamoda2.5 achieves up to $95.9\times$ faster video editing inference. In real-world applications, Mamoda2.5 has been successfully deployed for content moderation and creative restoration tasks in advertising scenarios, achieving a 98% success rate in internal advertising video editing scenario.

Video diffusion models leveraging step distillation or causal distillation have achieved remarkable performance. However, adapting existing LoRAs to these variants remains a critical challenge due to weight space mismatches. We observe that direct application leads to style degradation and structural collapse, yet the underlying mechanisms remain poorly understood. To fill this gap, we delve into the weight space and identify that the incompatibility stems from spectral interference within shared functional clusters defined over singular subspaces. Specifically, our analysis reveals that while both paradigms respect spectral rigidity, they establish conflicting routing pathways that clash through constructive overload or destructive cancellation. To address this issue, we propose Cluster-Aware Spectral Arbitration (CASA), a data-free framework that dynamically arbitrates between safeguarding the target's manifold and restoring LoRA alignment based on spectral density. Extensive experiments demonstrate that CASA effectively mitigates artifacts and revives LoRA functionality. Our code is available at https://github.com/Noahwangyuchen/CASA

Embodied agents require robust and comprehensive 3D spatiotemporal representations to support spatial reasoning, manipulation understanding, and downstream decision making. However, existing robot data are typically captured from fixed or sparse viewpoints, providing only partial and view-dependent observations, which limits multi-view perception and generalization across viewpoints. Given the difficulty of collecting additional viewpoints in real-world settings, we propose Embody4D, a dedicated video-to-video world model for embodied scenarios to bridge this observation gap by transforming a monocular robot video into novel-view videos from flexible target camera viewpoints. First, to tackle training data scarcity, we introduce a 3D-aware compositional synthesis pipeline to curate a heterogeneous dataset compositing cross-embodiment robotic arms with diverse backgrounds, promoting broad generalization. Second, to enforce geometric stability, we devise a latent confidence-aware expert modulation strategy, which estimates the reliability of warped latent priors and adaptively routes regions to copy, repair, or inpaint experts for spatiotemporally consistent 4D generation. Finally, to enhance the fidelity of the manipulation, we incorporate an interaction-aware attention mechanism that explicitly attends to the robotic interaction regions. Extensive experiments show that Embody4D achieves state-of-the-art performance on visual evaluation benchmarks, while both simulated and real-world robotic experiments further demonstrate its effectiveness as a robust data engine for synthesizing high-fidelity, view-consistent videos that empower downstream robotic planning and learning.

Text-to-Video (T2V) models have demonstrated remarkable capability in generating temporally coherent videos from natural language prompts, yet they also risk producing unsafe content such as violence or explicit material. Existing prompt-level defenses are largely inherited from text-to-image safety and operate on the lexical surface of the input, making them vulnerable to jailbreak attacks that disguise harmful intent through rephrasing or adversarial prompting. Moreover, T2V generation introduces a distinctive challenge overlooked by prior work: temporally emergent risk, where a seemingly benign prompt leads to unsafe content through the generator's temporal extrapolation toward narrative coherence. We propose \method{}, a training-free, inference-time defense framework that reformulates T2V safety as a causal intervention in a temporally structured semantic space. TrajShield handles explicit unsafe prompts, jailbreak attacks, and temporally emergent risks in a unified manner by simulating the implied trajectory of a prompt, localizing the causal origin of potential risk, and applying a minimally invasive rewrite that neutralizes the risk while preserving safety-irrelevant semantics. Experiments on T2VSafetyBench across 14 safety categories and multiple T2V backends demonstrate that TrajShield achieves state-of-the-art defenseive performance while maintaining high semantic fidelity, substantially outperforming existing defenses, with an average ASR reduction of 52.44%.

Autoregressive video generation paradigms offer theoretical promise for long video synthesis, yet their practical deployment is hindered by the computational burden of sequential iterative denoising. While cache reuse strategies can accelerate generation by skipping redundant denoising steps, existing methods rely on coarse-grained chunk-level skipping that fails to capture fine-grained pixel dynamics. This oversight is critical: pixels with high motion require more denoising steps to prevent error accumulation, while static pixels tolerate aggressive skipping. We formalize this insight theoretically by linking cache errors to residual instability, and propose MotionCache, a motion-aware cache framework that exploits inter-frame differences as a lightweight proxy for pixel-level motion characteristics. MotionCache employs a coarse-to-fine strategy: an initial warm-up phase establishes semantic coherence, followed by motion-weighted cache reuse that dynamically adjusts update frequencies per token. Extensive experiments on state-of-the-art models like SkyReels-V2 and MAGI-1 demonstrate that MotionCache achieves significant speedups of $\textbf{6.28}\times$ and $\textbf{1.64}\times$ respectively, while effectively preserving generation quality (VBench: $1\%\downarrow$ and $0.01\%\downarrow$ respectively). The code is available at https://github.com/ywlq/MotionCache.

Recent progress has shown that video diffusion models (VDMs) can be repurposed for diverse multimodal graphics tasks. However, existing methods often train separate models for each problem setting, which fixes the input-output mapping and limits the modeling of correlations across modalities. We present UniVidX, a unified multimodal framework that leverages VDM priors for versatile video generation. UniVidX formulates pixel-aligned tasks as conditional generation in a shared multimodal space, adapts to modality-specific distributions while preserving the backbone's native priors, and promotes cross-modal consistency during synthesis. It is built on three key designs. Stochastic Condition Masking (SCM) randomly partitions modalities into clean conditions and noisy targets during training, enabling omni-directional conditional generation instead of fixed mappings. Decoupled Gated LoRA (DGL) introduces per-modality LoRAs that are activated when a modality serves as the generation target, preserving the strong priors of the VDM. Cross-Modal Self-Attention (CMSA) shares keys and values across modalities while keeping modality-specific queries, facilitating information exchange and inter-modal alignment. We instantiate UniVidX in two domains: UniVid-Intrinsic, for RGB videos and intrinsic maps including albedo, irradiance, and normal; and UniVid-Alpha, for blended RGB videos and their constituent RGBA layers. Experiments show that both models achieve performance competitive with state-of-the-art methods across distinct tasks and generalize robustly to in-the-wild scenarios, even when trained on fewer than 1,000 videos. Project page: https://houyuanchen111.github.io/UniVidX.github.io/

The proliferation of advanced AI video synthesis techniques poses an unprecedented challenge to digital video authenticity. Existing AI-generated video (AIGV) detection methods primarily focus on uni-modal or spatiotemporal artifacts, but they overlook the rich cues within the visual-textual cross-modal space, especially the temporal stability of semantic alignment. In this work, we identify a distinctive fingerprint in AIGVs, termed cross-modal temporal artifact (CMTA). Unlike real videos that exhibit natural temporal fluctuations in cross-modal alignment due to semantic variations, AIGVs display unnaturally stable semantic trajectories governed by given input prompts. To bridge this gap, we propose the CMTA framework, a cross-modal detection approach that captures these unique temporal artifacts through joint cross-modal embedding and multi-grained temporal modeling. Specifically, CMTA leverages BLIP to generate frame-level image captions and utilizes CLIP to extract corresponding visual-textual representations. A coarse-grained temporal modeling branch is then designed to characterize temporal fluctuations in cross-modal alignment with a GRU. In parallel, a fine-grained branch is constructed to capture intricate inter-frame variations from integrated visual-textual features with a Transformer encoder. Extensive experiments on 40 subsets across four large-scale datasets, including GenVideo, EvalCrafter, VideoPhy, and VidProM, validate that our approach sets a new state-of-the-art while exhibiting superior cross-generator generalization. Code and models of CMTA will be released at https://github.com/hwang-cs-ime/CMTA

Autoregressive (AR) models with diffusion heads have recently achieved strong text-to-audio performance, yet their iterative decoding and multi-step sampling process introduce high-latency issues. To address this bottleneck, we propose a one-step sampling framework that combines an energy-distance training objective with representation-level distillation. An energy-scoring head maps Gaussian noise directly to audio latents in one step, eliminating the need for a costly recursive diffusion sampling process, while distillation from a masked autoregressive (MAR) text-to-audio model preserves the strong conditioning learned during diffusion training. On the AudioCaps benchmark, our method consistently outperforms prior one-step baselines such as ConsistencyTTA, SoundCTM, AudioLCM and AudioTurbo, on both objective and subjective metrics, while substantially narrowing the quality gap to AR diffusion systems with multi-step sampling. Compared to the state-of-the-art AR diffusion system, IMPACT, our approach achieves up to $8.5$x faster batch inference with highly competitive audio quality. These results demonstrate that combining energy-distance training with representation-level distillation provides an effective recipe for fast, high-quality text-to-audio synthesis.

We introduce Lucid-XR, a generative data engine for creating diverse and realistic-looking multi-modal data to train real-world robotic systems. At the core of Lucid-XR is vuer, a web-based physics simulation environment that runs directly on the XR headset, enabling internet-scale access to immersive, latency-free virtual interactions without requiring specialized equipment. The complete system integrates on-device physics simulation with human-to-robot pose retargeting. Data collected is further amplified by a physics-guided video generation pipeline steerable via natural language specifications. We demonstrate zero-shot transfer of robot visual policies to unseen, cluttered, and badly lit evaluation environments, after training entirely on Lucid-XR's synthetic data. We include examples across dexterous manipulation tasks that involve soft materials, loosely bound particles, and rigid body contact. Project website: https://lucidxr.github.io

Recent visual generation models have made major progress in photorealism, typography, instruction following, and interactive editing, yet they still struggle with spatial reasoning, persistent state, long-horizon consistency, and causal understanding. We argue that the field should move beyond appearance synthesis toward intelligent visual generation: plausible visuals grounded in structure, dynamics, domain knowledge, and causal relations. To frame this shift, we introduce a five-level taxonomy: Atomic Generation, Conditional Generation, In-Context Generation, Agentic Generation, and World-Modeling Generation, progressing from passive renderers to interactive, agentic, world-aware generators. We analyze key technical drivers, including flow matching, unified understanding-and-generation models, improved visual representations, post-training, reward modeling, data curation, synthetic data distillation, and sampling acceleration. We further show that current evaluations often overestimate progress by emphasizing perceptual quality while missing structural, temporal, and causal failures. By combining benchmark review, in-the-wild stress tests, and expert-constrained case studies, this roadmap offers a capability-centered lens for understanding, evaluating, and advancing the next generation of intelligent visual generation systems.

Despite rapid advances in photorealistic video generation, real-world applications such as filmmaking require video aesthetics, e.g., harmonious colors and cinematic lighting, beyond visual fidelity. Prior work on visual aesthetics largely focuses on images, often reducing aesthetics to coarse definitions, e.g., visual pleasure, without a rigorous and systematic evaluation. To improve video aesthetics, we propose a hierarchical rubric that decomposes video aesthetics into three core dimensions, Visual Aesthetics (VA), Visual Fidelity (VF), and Visual Plausibility (VP), with 15 fine-grained criteria, e.g., shot composition. This framework enables a large-scale expert-annotated preference dataset and an evaluation benchmark, AesVideo-Bench, containing about 2500 video pairs with expert annotations on VA, VF, and VP. We then build a family of Video Aesthetic Reward Models (AesRM): AesRM-Base, which directly predicts pairwise preferences on these dimensions to provide efficient post-training rewards, and AesRM-CoT, which additionally generates CoT aligned with all 15 criteria to improve assessment interpretability. Specifically, we train AesRM with a three-stage progressive scheme: (1) Atomic Aesthetic Capability Learning, which strengthens AesRM's recognition of fundamental aesthetic concepts, e.g., accurately identifying centered composition; (2) Cold-Start, aligning the model with structured reasoning protocols; and (3) GRPO, further improving evaluation accuracy. To enhance AesRM-CoT, we additionally propose self-consistency-based CoT synthesis to improve CoT quality and design CoT-based process rewards during GRPO. Extensive experiments show AesRM outperforms baselines on multiple aesthetics benchmarks and is more robust, with lower position bias. Finally, we align Wan2.2 with AesRM and observe clear aesthetic gains over existing aesthetic reward models.

Humanoid control systems have made significant progress in recent years, yet modeling fluent interaction-rich behavior between a robot, its surrounding environment, and task-relevant objects remains a fundamental challenge. This difficulty arises from the need to jointly capture spatial context, temporal dynamics, robot actions, and task intent at scale, which is a poor match to conventional supervision. We propose ExoActor, a novel framework that leverages the generalization capabilities of large-scale video generation models to address this problem. The key insight in ExoActor is to use third-person video generation as a unified interface for modeling interaction dynamics. Given a task instruction and scene context, ExoActor synthesizes plausible execution processes that implicitly encode coordinated interactions between robot, environment, and objects. Such video output is then transformed into executable humanoid behaviors through a pipeline that estimates human motion and executes it via a general motion controller, yielding a task-conditioned behavior sequence. To validate the proposed framework, we implement it as an end-to-end system and demonstrate its generalization to new scenarios without additional real-world data collection. Furthermore, we conclude by discussing limitations of the current implementation and outlining promising directions for future research, illustrating how ExoActor provides a scalable approach to modeling interaction-rich humanoid behaviors, potentially opening a new avenue for generative models to advance general-purpose humanoid intelligence.

Generating continuous-time, continuous-space stochastic processes (e.g., videos, weather forecasts) conditioned on partial observations (e.g., first and last frames) is a fundamental challenge. Existing approaches, (e.g., diffusion models), suffer from key limitations: (1) noise-to-data evolution fails to capture structural similarity between states close in physical time and has unstable integration in low-step regimes; (2) random noise injected is insensitive to the physical process's time elapsed, resulting in incorrect dynamics; (3) they overlook conditioning on arbitrary subsets of states (e.g., irregularly sampled timesteps, future observations). We propose ABC: Any-Subset Autoregressive Models via Non-Markovian Diffusion Bridges in Continuous Time and Space. Crucially, we model the process with one continual SDE whose time variable and intermediate states track the real time and process states. This has provable advantages: (1) the starting point for generating future states is the already-close previous state, rather than uninformative noise; (2) random noise injection scales with physical time elapsed, encouraging physically plausible dynamics with similar time-adjacent states. We derive SDE dynamics via changes-of-measure on path space, yielding another advantage: (3) path-dependent conditioning on arbitrary subsets of the state history and/or future. To learn these dynamics, we derive a path- and time-dependent extension of denoising score matching. Our experiments show ABC's superiority to competing methods on multiple domains, including video generation and weather forecasting.

Recent advances in Diffusion Transformer (DiT)-based video generation technologies have shown impressive results for video object removal. However, these methods still suffer from substantial inference latency. For instance, although MiniMax Remover achieves state-of-the-art visual quality, it operates at only around 10FPS, primarily due to dense computations over the entire spatiotemporal token space, even when only a small masked region actually requires processing. In this paper, we present YOSE, You Only Select Essential Tokens, an efficient fine-tuning framework. YOSE introduces two key components: Batch Variable-length Indexing (BVI) and Diffusion Process Simulator (DiffSim) Module. BVI is a differentiable dynamic indexing operator that adaptively selects essential tokens based on mask information, enabling variable-length token processing across samples. DiffSim provides a diffusion process approximation mechanism for unmasked tokens, which simulates the influence of unmasked regions within DiT self-attention to maintain semantic consistency for masked tokens. With these designs, YOSE achieves mask-aware acceleration, where the inference time scales approximately linearly with the masked regions, in contrast to full-token diffusion methods whose computation remains constant regardless of the mask size. Extensive experiments demonstrate that YOSE achieves up to 2.5X speedup in 70% of cases while maintaining visual quality comparable to the baseline. Code is available at: https://github.com/Wucy0519/YOSE-CVPR26.

While large-scale video diffusion models have demonstrated impressive capabilities in generating high-resolution and semantically rich content, a significant gap remains between their pretraining performance and real-world deployment requirements due to critical issues such as prompt sensitivity, temporal inconsistency, and prohibitive inference costs. To bridge this gap, we propose a comprehensive post-training framework that systematically aligns pretrained models with user intentions through four synergistic stages: we first employ Supervised Fine-Tuning (SFT) to transform the base model into a stable instruction-following policy, followed by a Reinforcement Learning from Human Feedback (RLHF) stage that utilizes a novel Group Relative Policy Optimization (GRPO) method tailored for video diffusion to enhance perceptual quality and temporal coherence; subsequently, we integrate Prompt Enhancement via a specialized language model to refine user inputs, and finally address system efficiency through Inference Optimization. Together, these components provide a systematic approach to improving visual quality, temporal coherence, and instruction following, while preserving the controllability learned during pretraining. The result is a practical blueprint for building scalable post-training pipelines that are stable, adaptable, and effective in real-world deployment. Extensive experiments demonstrate that this unified pipeline effectively mitigates common artifacts and significantly improves controllability and visual aesthetics while adhering to strict sampling cost constraints.