Collecting multi-view driving scenario videos to enhance the performance of 3D visual perception tasks presents significant challenges and incurs substantial costs, making generative models for realistic data an appealing alternative. Yet, the videos generated by recent works suffer from poor quality and spatiotemporal consistency, undermining their utility in advancing perception tasks under driving scenarios. To address this gap, we propose DiVE, a diffusion transformer-based generative framework meticulously engineered to produce high-fidelity, temporally coherent, and cross-view consistent multi-view videos, aligning seamlessly with bird's-eye view layouts and textual descriptions. DiVE leverages a unified cross-attention and a SketchFormer to exert precise control over multimodal data, while incorporating a view-inflated attention mechanism that adds no extra parameters, thereby guaranteeing consistency across views. Despite these advancements, synthesizing high-resolution videos under multimodal constraints introduces dual challenges: investigating the optimal classifier-free guidance coniguration under intricate multi-condition inputs and mitigating excessive computational latency in high-resolution rendering--both of which remain underexplored in prior researches. To resolve these limitations, we introduce two innovations: Multi-Control Auxiliary Branch Distillation, which streamlines multi-condition CFG selection while circumventing high computational overhead, and Resolution Progressive Sampling, a training-free acceleration strategy that staggers resolution scaling to reduce high latency due to high resolution. These innovations collectively achieve a 2.62x speedup with minimal quality degradation. Evaluated on the nuScenes dataset, DiVE achieves SOTA performance in multi-view video generation, yielding photorealistic outputs with exceptional temporal and cross-view coherence.
Current text-to-video (T2V) generation models are increasingly popular due to their ability to produce coherent videos from textual prompts. However, these models often struggle to generate semantically and temporally consistent videos when dealing with longer, more complex prompts involving multiple objects or sequential events. Additionally, the high computational costs associated with training or fine-tuning make direct improvements impractical. To overcome these limitations, we introduce NeuS-E, a novel zero-training video refinement pipeline that leverages neuro-symbolic feedback to automatically enhance video generation, achieving superior alignment with the prompts. Our approach first derives the neuro-symbolic feedback by analyzing a formal video representation and pinpoints semantically inconsistent events, objects, and their corresponding frames. This feedback then guides targeted edits to the original video. Extensive empirical evaluations on both open-source and proprietary T2V models demonstrate that NeuS-E significantly enhances temporal and logical alignment across diverse prompts by almost 40%
While recent advancements in robotic manipulation video synthesis have shown promise, significant challenges persist in ensuring effective instruction-following and achieving high visual quality. Recent methods, like RoboDreamer, utilize linguistic decomposition to divide instructions into separate lower-level primitives, conditioning the world model on these primitives to achieve compositional instruction-following. However, these separate primitives do not consider the relationships that exist between them. Furthermore, recent methods neglect valuable visual guidance, including depth and semantic guidance, both crucial for enhancing visual quality. This paper introduces ManipDreamer, an advanced world model based on the action tree and visual guidance. To better learn the relationships between instruction primitives, we represent the instruction as the action tree and assign embeddings to tree nodes, each instruction can acquire its embeddings by navigating through the action tree. The instruction embeddings can be used to guide the world model. To enhance visual quality, we combine depth and semantic guidance by introducing a visual guidance adapter compatible with the world model. This visual adapter enhances both the temporal and physical consistency of video generation. Based on the action tree and visual guidance, ManipDreamer significantly boosts the instruction-following ability and visual quality. Comprehensive evaluations on robotic manipulation benchmarks reveal that ManipDreamer achieves large improvements in video quality metrics in both seen and unseen tasks, with PSNR improved from 19.55 to 21.05, SSIM improved from 0.7474 to 0.7982 and reduced Flow Error from 3.506 to 3.201 in unseen tasks, compared to the recent RoboDreamer model. Additionally, our method increases the success rate of robotic manipulation tasks by 2.5% in 6 RLbench tasks on average.
Despite recent progress in video generation, producing videos that adhere to physical laws remains a significant challenge. Traditional diffusion-based methods struggle to extrapolate to unseen physical conditions (eg, velocity) due to their reliance on data-driven approximations. To address this, we propose to integrate symbolic reasoning and reinforcement learning to enforce physical consistency in video generation. We first introduce the Diffusion Timestep Tokenizer (DDT), which learns discrete, recursive visual tokens by recovering visual attributes lost during the diffusion process. The recursive visual tokens enable symbolic reasoning by a large language model. Based on it, we propose the Phys-AR framework, which consists of two stages: The first stage uses supervised fine-tuning to transfer symbolic knowledge, while the second stage applies reinforcement learning to optimize the model's reasoning abilities through reward functions based on physical conditions. Our approach allows the model to dynamically adjust and improve the physical properties of generated videos, ensuring adherence to physical laws. Experimental results demonstrate that PhysAR can generate videos that are physically consistent.
Classifier-free guidance (CFG) is a key technique for improving conditional generation in diffusion models, enabling more accurate control while enhancing sample quality. It is natural to extend this technique to video diffusion, which generates video conditioned on a variable number of context frames, collectively referred to as history. However, we find two key challenges to guiding with variable-length history: architectures that only support fixed-size conditioning, and the empirical observation that CFG-style history dropout performs poorly. To address this, we propose the Diffusion Forcing Transformer (DFoT), a video diffusion architecture and theoretically grounded training objective that jointly enable conditioning on a flexible number of history frames. We then introduce History Guidance, a family of guidance methods uniquely enabled by DFoT. We show that its simplest form, vanilla history guidance, already significantly improves video generation quality and temporal consistency. A more advanced method, history guidance across time and frequency further enhances motion dynamics, enables compositional generalization to out-of-distribution history, and can stably roll out extremely long videos. Website: https://boyuan.space/history-guidance
We propose to train a subject-driven customized video generation model through decoupling the subject-specific learning from temporal dynamics in zero-shot without additional tuning. A traditional method for video customization that is tuning-free often relies on large, annotated video datasets, which are computationally expensive and require extensive annotation. In contrast to the previous approach, we introduce the use of an image customization dataset directly on training video customization models, factorizing the video customization into two folds: (1) identity injection through image customization dataset and (2) temporal modeling preservation with a small set of unannotated videos through the image-to-video training method. Additionally, we employ random image token dropping with randomized image initialization during image-to-video fine-tuning to mitigate the copy-and-paste issue. To further enhance learning, we introduce stochastic switching during joint optimization of subject-specific and temporal features, mitigating catastrophic forgetting. Our method achieves strong subject consistency and scalability, outperforming existing video customization models in zero-shot settings, demonstrating the effectiveness of our framework.
Diffusion models have demonstrated appealing performance in both image and video generation. However, many works discover that they struggle to capture important, high-level relationships that are present in the real world. For example, they fail to learn physical laws from data, and even fail to understand that the objects in the world exist in a stable fashion. This is due to the fact that important conditional dependence structures are not adequately captured in the vanilla diffusion models. In this work, we initiate an in-depth study on strengthening the diffusion model to capture the conditional dependence structures in the data. In particular, we examine the efficacy of the auto-regressive (AR) diffusion models for such purpose and develop the first theoretical results on the sampling error of AR diffusion models under (possibly) the mildest data assumption. Our theoretical findings indicate that, compared with typical diffusion models, the AR variant produces samples with a reduced gap in approximating the data conditional distribution. On the other hand, the overall inference time of the AR-diffusion models is only moderately larger than that for the vanilla diffusion models, making them still practical for large scale applications. We also provide empirical results showing that when there is clear conditional dependence structure in the data, the AR diffusion models captures such structure, whereas vanilla DDPM fails to do so. On the other hand, when there is no obvious conditional dependence across patches of the data, AR diffusion does not outperform DDPM.
This paper presents an effective approach for learning novel 4D embodied world models, which predict the dynamic evolution of 3D scenes over time in response to an embodied agent's actions, providing both spatial and temporal consistency. We propose to learn a 4D world model by training on RGB-DN (RGB, Depth, and Normal) videos. This not only surpasses traditional 2D models by incorporating detailed shape, configuration, and temporal changes into their predictions, but also allows us to effectively learn accurate inverse dynamic models for an embodied agent. Specifically, we first extend existing robotic manipulation video datasets with depth and normal information leveraging off-the-shelf models. Next, we fine-tune a video generation model on this annotated dataset, which jointly predicts RGB-DN (RGB, Depth, and Normal) for each frame. We then present an algorithm to directly convert generated RGB, Depth, and Normal videos into a high-quality 4D scene of the world. Our method ensures temporal and spatial coherence in 4D scene predictions from embodied scenarios, enables novel view synthesis for embodied environments, and facilitates policy learning that significantly outperforms those derived from prior video-based world models.
Text-to-video generative models have made significant strides in recent years, producing high-quality videos that excel in both aesthetic appeal and accurate instruction following, and have become central to digital art creation and user engagement online. Yet, despite these advancements, their ability to respect fundamental physical laws remains largely untested: many outputs still violate basic constraints such as rigid-body collisions, energy conservation, and gravitational dynamics, resulting in unrealistic or even misleading content. Existing physical-evaluation benchmarks typically rely on automatic, pixel-level metrics applied to simplistic, life-scenario prompts, and thus overlook both human judgment and first-principles physics. To fill this gap, we introduce \textbf{T2VPhysBench}, a first-principled benchmark that systematically evaluates whether state-of-the-art text-to-video systems, both open-source and commercial, obey twelve core physical laws including Newtonian mechanics, conservation principles, and phenomenological effects. Our benchmark employs a rigorous human evaluation protocol and includes three targeted studies: (1) an overall compliance assessment showing that all models score below 0.60 on average in each law category; (2) a prompt-hint ablation revealing that even detailed, law-specific hints fail to remedy physics violations; and (3) a counterfactual robustness test demonstrating that models often generate videos that explicitly break physical rules when so instructed. The results expose persistent limitations in current architectures and offer concrete insights for guiding future research toward truly physics-aware video generation.
Transformers serve as the foundational architecture for large language and video generation models, such as GPT, BERT, SORA and their successors. Empirical studies have demonstrated that real-world data and learning tasks exhibit low-dimensional structures, along with some noise or measurement error. The performance of transformers tends to depend on the intrinsic dimension of the data/tasks, though theoretical understandings remain largely unexplored for transformers. This work establishes a theoretical foundation by analyzing the performance of transformers for regression tasks involving noisy input data on a manifold. Specifically, the input data are in a tubular neighborhood of a manifold, while the ground truth function depends on the projection of the noisy data onto the manifold. We prove approximation and generalization errors which crucially depend on the intrinsic dimension of the manifold. Our results demonstrate that transformers can leverage low-complexity structures in learning task even when the input data are perturbed by high-dimensional noise. Our novel proof technique constructs representations of basic arithmetic operations by transformers, which may hold independent interest.
Recent advancements in autonomous driving (AD) systems have highlighted the potential of world models in achieving robust and generalizable performance across both ordinary and challenging driving conditions. However, a key challenge remains: precise and flexible camera pose control, which is crucial for accurate viewpoint transformation and realistic simulation of scene dynamics. In this paper, we introduce PosePilot, a lightweight yet powerful framework that significantly enhances camera pose controllability in generative world models. Drawing inspiration from self-supervised depth estimation, PosePilot leverages structure-from-motion principles to establish a tight coupling between camera pose and video generation. Specifically, we incorporate self-supervised depth and pose readouts, allowing the model to infer depth and relative camera motion directly from video sequences. These outputs drive pose-aware frame warping, guided by a photometric warping loss that enforces geometric consistency across synthesized frames. To further refine camera pose estimation, we introduce a reverse warping step and a pose regression loss, improving viewpoint precision and adaptability. Extensive experiments on autonomous driving and general-domain video datasets demonstrate that PosePilot significantly enhances structural understanding and motion reasoning in both diffusion-based and auto-regressive world models. By steering camera pose with self-supervised depth, PosePilot sets a new benchmark for pose controllability, enabling physically consistent, reliable viewpoint synthesis in generative world models.
Synthetic video generation has gained significant attention for its realism and broad applications, but remains prone to violations of common sense and physical laws. This highlights the need for reliable abnormality detectors that understand such principles and are robust to hallucinations. To address this, we introduce VideoHallu, a benchmark of over 3,000 video QA pairs built from synthetic videos generated by models like Veo2, Sora, and Kling, paired with expert-crafted counterintuitive QA to evaluate the critical thinking abilities of Multi-modal Large Language Models (MLLMs) on abnormalities that are perceptually obvious to humans but often hallucinated due to language priors. VideoHallu evaluates MLLMs' abnormality detection abilities with examples across alignment, consistency, commonsense, and physics. We benchmark SOTA MLLMs, including GPT-4o, Gemini-2.5-Pro, Qwen2.5-VL, Video-R1, and VideoChat-R1. We observe that these models perform well on many real-world benchmarks like MVBench and MovieChat, but still struggle with basic physics-based and commonsense reasoning in synthetic videos. We further show that post-training with Group Relative Policy Optimization (GRPO), using curriculum learning on datasets combining video QA with counterintuitive commonsense and physics reasoning over real and synthetic videos, improves MLLMs' abnormality detection and critical thinking, demonstrating the value of targeted training for improving their understanding of commonsense and physical laws.
Long video generation involves generating extended videos using models trained on short videos, suffering from distribution shifts due to varying frame counts. It necessitates the use of local information from the original short frames to enhance visual and motion quality, and global information from the entire long frames to ensure appearance consistency. Existing training-free methods struggle to effectively integrate the benefits of both, as appearance and motion in videos are closely coupled, leading to motion inconsistency and visual quality. In this paper, we reveal that global and local information can be precisely decoupled into consistent appearance and motion intensity information by applying Principal Component Analysis (PCA), allowing for refined complementary integration of global consistency and local quality. With this insight, we propose FreePCA, a training-free long video generation paradigm based on PCA that simultaneously achieves high consistency and quality. Concretely, we decouple consistent appearance and motion intensity features by measuring cosine similarity in the principal component space. Critically, we progressively integrate these features to preserve original quality and ensure smooth transitions, while further enhancing consistency by reusing the mean statistics of the initial noise. Experiments demonstrate that FreePCA can be applied to various video diffusion models without requiring training, leading to substantial improvements. Code is available at https://github.com/JosephTiTan/FreePCA.
Audio-driven human animation technology is widely used in human-computer interaction, and the emergence of diffusion models has further advanced its development. Currently, most methods rely on multi-stage generation and intermediate representations, resulting in long inference time and issues with generation quality in specific foreground regions and audio-motion consistency. These shortcomings are primarily due to the lack of localized fine-grained supervised guidance. To address above challenges, we propose Parts-aware Audio-driven Human Animation, PAHA, a unit enhancement and guidance framework for audio-driven upper-body animation. We introduce two key methods: Parts-Aware Re-weighting (PAR) and Parts Consistency Enhancement (PCE). PAR dynamically adjusts regional training loss weights based on pose confidence scores, effectively improving visual quality. PCE constructs and trains diffusion-based regional audio-visual classifiers to improve the consistency of motion and co-speech audio. Afterwards, we design two novel inference guidance methods for the foregoing classifiers, Sequential Guidance (SG) and Differential Guidance (DG), to balance efficiency and quality respectively. Additionally, we build CNAS, the first public Chinese News Anchor Speech dataset, to advance research and validation in this field. Extensive experimental results and user studies demonstrate that PAHA significantly outperforms existing methods in audio-motion alignment and video-related evaluations. The codes and CNAS dataset will be released upon acceptance.
Customized text-to-video generation with pre-trained large-scale models has recently garnered significant attention through focusing on identity and motion consistency. Existing works typically follow the isolated customized paradigm, where the subject identity or motion dynamics are customized exclusively. However, this paradigm completely ignores the intrinsic mutual constraints and synergistic interdependencies between identity and motion, resulting in identity-motion conflicts throughout the generation process that systematically degrades. To address this, we introduce DualReal, a novel framework that, employs adaptive joint training to collaboratively construct interdependencies between dimensions. Specifically, DualReal is composed of two units: (1) Dual-aware Adaptation dynamically selects a training phase (i.e., identity or motion), learns the current information guided by the frozen dimension prior, and employs a regularization strategy to avoid knowledge leakage; (2) StageBlender Controller leverages the denoising stages and Diffusion Transformer depths to guide different dimensions with adaptive granularity, avoiding conflicts at various stages and ultimately achieving lossless fusion of identity and motion patterns. We constructed a more comprehensive benchmark than existing methods. The experimental results show that DualReal improves CLIP-I and DINO-I metrics by 21.7% and 31.8% on average, and achieves top performance on nearly all motion quality metrics.
Pose-Guided Person Image Synthesis (PGPIS) generates realistic person images conditioned on a target pose and a source image. This task plays a key role in various real-world applications, such as sign language video generation, AR/VR, gaming, and live streaming. In these scenarios, real-time PGPIS is critical for providing immediate visual feedback and maintaining user immersion.However, achieving real-time performance remains a significant challenge due to the complexity of synthesizing high-fidelity images from diverse and dynamic human poses. Recent diffusion-based methods have shown impressive image quality in PGPIS, but their slow sampling speeds hinder deployment in time-sensitive applications. This latency is particularly problematic in tasks like generating sign language videos during live broadcasts, where rapid image updates are required. Therefore, developing a fast and reliable PGPIS model is a crucial step toward enabling real-time interactive systems. To address this challenge, we propose a generative model based on flow matching (FM). Our approach enables faster, more stable, and more efficient training and sampling. Furthermore, the proposed model supports conditional generation and can operate in latent space, making it especially suitable for real-time PGPIS applications where both speed and quality are critical. We evaluate our proposed method, Real-Time Person Image Synthesis Using a Flow Matching Model (RPFM), on the widely used DeepFashion dataset for PGPIS tasks. Our results show that RPFM achieves near-real-time sampling speeds while maintaining performance comparable to the state-of-the-art models. Our methodology trades off a slight, acceptable decrease in generated-image accuracy for over a twofold increase in generation speed, thereby ensuring real-time performance.
Demand for 2K video synthesis is rising with increasing consumer expectations for ultra-clear visuals. While diffusion transformers (DiTs) have demonstrated remarkable capabilities in high-quality video generation, scaling them to 2K resolution remains computationally prohibitive due to quadratic growth in memory and processing costs. In this work, we propose Turbo2K, an efficient and practical framework for generating detail-rich 2K videos while significantly improving training and inference efficiency. First, Turbo2K operates in a highly compressed latent space, reducing computational complexity and memory footprint, making high-resolution video synthesis feasible. However, the high compression ratio of the VAE and limited model size impose constraints on generative quality. To mitigate this, we introduce a knowledge distillation strategy that enables a smaller student model to inherit the generative capacity of a larger, more powerful teacher model. Our analysis reveals that, despite differences in latent spaces and architectures, DiTs exhibit structural similarities in their internal representations, facilitating effective knowledge transfer. Second, we design a hierarchical two-stage synthesis framework that first generates multi-level feature at lower resolutions before guiding high-resolution video generation. This approach ensures structural coherence and fine-grained detail refinement while eliminating redundant encoding-decoding overhead, further enhancing computational efficiency.Turbo2K achieves state-of-the-art efficiency, generating 5-second, 24fps, 2K videos with significantly reduced computational cost. Compared to existing methods, Turbo2K is up to 20$\times$ faster for inference, making high-resolution video generation more scalable and practical for real-world applications.
The increasing demand for AR/VR applications has highlighted the need for high-quality 360-degree panoramic content. However, generating high-quality 360-degree panoramic images and videos remains a challenging task due to the severe distortions introduced by equirectangular projection (ERP). Existing approaches either fine-tune pretrained diffusion models on limited ERP datasets or attempt tuning-free methods that still rely on ERP latent representations, leading to discontinuities near the poles. In this paper, we introduce SphereDiff, a novel approach for seamless 360-degree panoramic image and video generation using state-of-the-art diffusion models without additional tuning. We define a spherical latent representation that ensures uniform distribution across all perspectives, mitigating the distortions inherent in ERP. We extend MultiDiffusion to spherical latent space and propose a spherical latent sampling method to enable direct use of pretrained diffusion models. Moreover, we introduce distortion-aware weighted averaging to further improve the generation quality in the projection process. Our method outperforms existing approaches in generating 360-degree panoramic content while maintaining high fidelity, making it a robust solution for immersive AR/VR applications. The code is available here. https://github.com/pmh9960/SphereDiff
The recent surge in open-source text-to-video generation models has significantly energized the research community, yet their dependence on proprietary training datasets remains a key constraint. While existing open datasets like Koala-36M employ algorithmic filtering of web-scraped videos from early platforms, they still lack the quality required for fine-tuning advanced video generation models. We present Tiger200K, a manually curated high visual quality video dataset sourced from User-Generated Content (UGC) platforms. By prioritizing visual fidelity and aesthetic quality, Tiger200K underscores the critical role of human expertise in data curation, and providing high-quality, temporally consistent video-text pairs for fine-tuning and optimizing video generation architectures through a simple but effective pipeline including shot boundary detection, OCR, border detecting, motion filter and fine bilingual caption. The dataset will undergo ongoing expansion and be released as an open-source initiative to advance research and applications in video generative models. Project page: https://tinytigerpan.github.io/tiger200k/
Compositional text-to-video generation, which requires synthesizing dynamic scenes with multiple interacting entities and precise spatial-temporal relationships, remains a critical challenge for diffusion-based models. Existing methods struggle with layout discontinuity, entity identity drift, and implausible interaction dynamics due to unconstrained cross-attention mechanisms and inadequate physics-aware reasoning. To address these limitations, we propose DyST-XL, a \textbf{training-free} framework that enhances off-the-shelf text-to-video models (e.g., CogVideoX-5B) through frame-aware control. DyST-XL integrates three key innovations: (1) A Dynamic Layout Planner that leverages large language models (LLMs) to parse input prompts into entity-attribute graphs and generates physics-aware keyframe layouts, with intermediate frames interpolated via trajectory optimization; (2) A Dual-Prompt Controlled Attention Mechanism that enforces localized text-video alignment through frame-aware attention masking, achieving the precise control over individual entities; and (3) An Entity-Consistency Constraint strategy that propagates first-frame feature embeddings to subsequent frames during denoising, preserving object identity without manual annotation. Experiments demonstrate that DyST-XL excels in compositional text-to-video generation, significantly improving performance on complex prompts and bridging a crucial gap in training-free video synthesis.