Pre-Trained Video Diffusion Models

Video diffusion transformers (DiTs) generate high-fidelity and temporally coherent videos, yet motion control remains implicit, primarily relying on text prompts. As a result, achieving desired motion often requires extensive prompt engineering and repeated resampling. While fine-tuning models with additional spatial prompts (e.g., bounding boxes or point trajectories) enables explicit control, it demands substantial data curation and computation, and may compromise the generative capabilities of pretrained models. Consequently, training-free motion control using such spatial prompts has been explored in U-Net-based video diffusion models, but remains largely unexplored for DiTs. We introduce QWERTY, a training-free framework that enables flexible motion control in pretrained image-to-video DiTs via user-defined object warping and optical flow. We carefully manipulate the 3D full attention of DiTs by warping the frame-invariant semantic subspace of queries. We find that the noise predicted by the query-warped DiT naturally guides the diffusion trajectory toward the desired motion, and further show that leveraging this noise as self-guidance for latent optimization improves control stability and visual quality. Experiments show that QWERTY achieves the most effective motion control among existing training-free approaches on a recent image-to-video DiT, with performance comparable to fine-tuning-based methods.

Monocular video depth estimation requires temporal consistency, geometric accuracy, and generalization across diverse scenarios, yet existing methods struggle to achieve all three simultaneously. Discriminative models excel at per-frame accuracy but suffer from temporal drift due to limited context windows, while generative methods improve consistency and generalization at the cost of extensive training data (10M+ samples) and lack of geometric precision. In response to these issues, we introduce \textbf{ICDepth}, a framework that adapts pre-trained text-to-video diffusion transformers for video depth estimation via In-Context Conditioning (ICC), leveraging their rich spatial-temporal priors. To address key challenges in transferring ICC from generation to dense prediction, we propose: (1)\textbf{SAND-Attention}, which ensures precise spatial-temporal alignment via shared RoPE and enforces unidirectional attention to prevent noise contamination; (2)\textbf{SRFM}, which injects DINOv2 semantic and resolution priors to enhance geometric precision. ICDepth achieves state-of-the-art results on multiple benchmarks with remarkable data efficiency, trained on only 0.8M frames ($6$--$13\times$ less than competing generative methods), while demonstrating strong zero-shot generalization to diverse domains.

How to accurately predict a high-fidelity future world? While the visual world is inherently continuous, existing deterministic video prediction models operate in discrete pixel space and are mainly optimized with pixel-wise mean squared error (MSE), which often leads to over-smoothed predictions and a lack of fine-grained visual details. To address these limitations, we propose Predictive Differentiable Rendering (PDR), a novel end-to-end video prediction paradigm that bridges the gap between discrete and continuous representations. Inspired by recent progress in 3D reconstruction with 3D Gaussian Splatting, we introduce PredGS, a lightweight and plug-and-play adapter based on 2D Gaussian representation, which could be seamlessly integrated with existing pixel space predictors, significantly improving spatial detail preservation with negligible computational overhead. Furthermore, we develop predgsplat, a CUDA-accelerated differentiable 2D Gaussian renderer supporting arbitrary channels. Each Gaussian is defined by 5 + C learnable parameters (position, scale, rotation, and C channel amplitudes) and achieves up to 10x faster rendering than the baseline. Optimized by a combined L1 and SSIM loss, PDR overcomes the inherent blurring tendencies of MSE Loss, significantly enhancing the prediction performance. Extensive experiments on diverse real-world benchmarks, including TaxiBJ, WeatherBench, KTH, and Human3.6M, demonstrate that PDR consistently surpasses existing methods, delivering superior detail preservation, visual fidelity, and predictive accuracy.

4D hand motion reconstruction from egocentric video is bottlenecked by clear limitations of existing methods: image-based pipelines depend on a detector that fails under heavy occlusion, while video-based methods rely on temporal modules learned only from scarce hand-pose annotations, a narrow signal insufficient to model motion dynamics, occlusion reasoning, and hand-object interaction. These capabilities, however, are exactly what video generative models must implicitly acquire when trained to synthesize coherent video at internet scale. Motivated by this, we present ViDiHand, which leverages the representations of a pretrained video diffusion model to reconstruct 4D two-hand pose. We adapt it via a hand-overlay rendering objective that specializes its features for hands while preserving its world priors. A decoder then recovers metric-scale pose from the adapted features. The whole pipeline operates directly on full frames--no detector, no infiller, and no test-time optimization. On ARCTIC, HOT3D, and HOI4D, ViDiHand substantially outperforms prior methods, establishing video diffusion models as a powerful new foundation for hand motion reconstruction and a promising route to scalable in-the-wild data collection for embodied AI. Project page: https://vidihand.github.io.

As the most perceptually powerful Face Video Super-Resolution (FVSR) method, existing works in Generative FVSR (GFVSR) mainly exploit the generative prior of pretrained diffusion models. However, viewed as full generation, they suffer from fixed sampling and expensive inference costs if without large-scale auxiliary training. Furthermore, an excessive pursuit of generic perceptual metrics often results in low fidelity. To address these issues, we present Dynamic Trajectory Initialization (DTI) paradigm for GFVSR, which reformulates GFVSR as an input-driven directional restoration. With a novel enhancement-and-injection conditioning mechanism for pretrained DiT backbone, fidelity of our model has been significantly improved without compromising perceptual quality. To dynamically set the starting sampling point, we propose a Discriminative Guide (DG) trained via objective Signal-to-Noise Ratio (SNR) alignment. With only minor model adaptation and fine-tuning, our method achieves a SOTA overall performance across diverse metrics and benchmarks. An analysis of relationship between actual comprehensive quality and common metrics is also conducted, which demonstrates the perception-distortion trade-off and that the LPIPS is the most convincing metric in our case.

Diffusion-based video relighting enables controllable relighting from a single input video, but modern video diffusion backbones are trained on short clips and applied to long-horizon videos through chunked sliding-window inference, often causing temporal discontinuities at chunk boundaries. We address this by reframing long-horizon relighting as \emph{temporally conditioned latent domain translation}. Our framework enforces cross-chunk continuity by propagating target-domain latents across boundaries and makes this behavior learnable using \emph{masked target-domain self-conditioning}, training the model to continue from temporally masked propagated context. We further introduce \emph{warm-start prompting} with a relit prompt anchor from a controllable generative model, which establishes the initial target-domain state and creates a general interface for prompt-based relighting. Experiments on in-the-wild long-horizon videos show markedly improved temporal consistency, with chunk-boundary artifacts largely reduced and unwanted appearance changes across chunks greatly suppressed.

Recent advances in video diffusion models have greatly improved visual fidelity, yet their generated motions often violate physical plausibility. We observe a common kinematic failure, "motion entanglement", the unintended coupling of independent motion sources, such as camera movement and object motion. We identify that this issue stems from data bias and the reconstruction-based training design of diffusion models. Training on noisy videos that still retain coarse motion cues inadvertently encourages the model to replicate existing motion without an incentive to learn how to model kinematically-grounded motions. To address this, we propose a Self-Imagination Fine-Tuning (SIFT) paradigm, which enables the model to learn from its own generated videos rather than directly reconstructing real ones, breaking the reconstruction shortcut. We further employ motion-aware discriminative supervision and a progressive hard-case replay strategy to stabilize and accelerate learning. By leveraging freely-generated text prompts, our method can densely cover a broad motion space, including rare or finely-disentangled scenarios that would be costly to collect as video data. Extensive experiments demonstrate that our approach substantially improves the physical realism, motion disentanglement, and controllability of generated videos.

Modern video diffusion transformers position their tokens through RoPE on the (u,v,t) axes -- a description of the camera's sampling grid that says nothing about the 3D structure of the scene. We observe that the geometric relation between two camera rays is captured by the Plucker reciprocal product, which is bilinear in the two rays -- the same algebraic form as the dot product in Transformer attention. Building on this analogy, we propose RayPE, a positional-encoding extension that injects per-token 6D Plucker coordinates additively into the queries and keys of self-attention, with a query/key flip arrangement under which the symmetric identity configuration coincides exactly with the reciprocal product. The injection is additive, the resulting attention score decomposes into a content term, a geometry term, and two content and geometry cross-terms -- all of which our experiments find individually necessary. To make the encoding stable across video data with heterogeneous camera-translation scales (SfM, deep SLAM, metric), we further decouple ray direction from moment magnitude, gate the encoding by a learned function of the log-magnitude, and apply RMSNorm to align it with the QKNorm-normalized content branch. The full module adds less than 0.1% parameters to a pretrained video DiT, is zero-initialized to start from the pretrained weights, and improves camera controllability, cross-frame 3D consistency, and overall video quality on a four-dataset training mixture.

While Video Virtual Try-on (VVT) has achieved remarkable progress in synthesizing realistic garment overlays on dynamic subjects, existing paradigms remains fundamentally constrained by a passive dependency on source camera trajectories, failing to accommodate the requisite interactive freedom for omnidirectional viewpoint exploration. To address this limitation, we define a pioneering research frontier: Camera-controllable Video Virtual Try-on (CaM-VVT). Unlike conventional VVT, CaM-VVT not only necessitates viewpoint-agnostic texture hallucination but also strict structural synchronization between non-rigid human dynamics and background contexts under arbitrary, unconstrained camera movements. To tackle these challenges, we present TryOnCrafter, the first unified DiT-based framework specifically architected for the CaM-VVT task. Departing from implicit pixel-space manipulation, we introduce a Renderable 4D Try-on Proxy that explicitly decouples the human subject from the environment. This is achieved by distilling high-fidelity 2D try-on priors into a clothed 3DGS-based avatar, which is subsequently animated via SMPL-X sequences and metric-aligned into a reconstructed background point cloud. This proxy establishes a robust structural foundation with superior texture density and motion integrity. Our Proxy-Anchored Video DiT leverages this robust structural foundation as a primary geometric anchor, ensuring that the synthesized photorealistic videos are strictly constrained by prescribed trajectories and physically plausible deformations. Benefiting from the inherent editability of the 4D proxy, TryOnCrafter facilitates diverse downstream applications, including human relocalization, ``bullet time'' effects, and $360$-degree orbital viewing.

Reconstructing 3D scenes from a single image is a fundamental challenge in computer vision, with broad applications in virtual reality, robotics, and content creation. Recent methods achieve outstanding performance by leveraging camera-controlled video diffusion models, but rely on iterative diffusion sampling, which greatly limits their practical deployment. We observe that geometric forward warping alone can cover the majority of a target view directly from the input image, with only a compact residual left for the encoder to correct. Motivated by this observation, we propose PRISM, a feed-forward framework that decomposes multi-view latent prediction into a parameter-free geometric prior and a learned residual correction, with no diffusion sampling required at inference. To enable generalization from purely synthetic training data, we devise a two-stage training strategy combining latents supervised distillation for geometric generalization and perceptual fine-tuning for appearance quality optimization. Extensive experiments on three benchmarks demonstrate that PRISM achieves competitive reconstruction quality compared with diffusion-based methods, while reducing inference time dramatically to only 36 seconds per scene.

While Deep Unfolding Networks (DUNs) dominate video Snapshot Compressive Imaging (SCI), they remain constrained by a uniform design philosophy. Existing methods repeatedly stack high-complexity priors with identical structures, ignoring the fact that optimization trajectories converge toward static states. This results in representation stagnation, where high-cost computations are wasted on minimal feature updates. To address this inefficiency, we present Differential Unfolding (DU), a heterogeneous framework that replaces uniform repetition with dynamic evolution. Central to DU is the Differential Evolutionary Framework (DEF), which partitions the unfolding process into two complementary roles: structural anchoring and differential evolution. In this scheme, high-parameter general stages are sparsely deployed to generate high-fidelity feature foundations. Complementing these, lightweight differential stages employ a Differential Representation Prior (DRP) to propagate and refine these foundational features through a differential mechanism. By integrating Differential Representation Attention (DRA) for evolving attention maps and a Differential Modulated FFN (DM-FFN) for feature rectification, DRP effectively models cross-stage variations with minimal overhead. By focusing computational resources on dynamic evolution rather than static redundancy, DU achieves a superior trade-off between accuracy and efficiency. Extensive experiments verify that our method establishes new state-of-the-art results while significantly slashing computational overhead. https://github.com/Muyuan-Zhang/DU

Video diffusion models have enabled remarkable progress in video generation and editing. However, content preservation remains a core challenge: existing methods regenerate every pixel and often alter elements that should remain unchanged, such as characters or background scenes. We introduce Vera, a layered diffusion framework for content-preserving video editing. Instead of regenerating the entire video, Vera generates an edit layer along with an alpha matte for compositing with the source video, separating creative editing from content preservation by design. To encourage coherent composition with the source video, we extend the text-to-video DiT into a Mixture-of-Transformers (MoT) architecture, with separate DiTs for each layer that interact through joint self-attention. To support the training of Vera, we further construct a high-quality layered dataset with accurate alpha mattes, diverse scenes and dynamics, and visual effects. Across our quantitative benchmark and human preference study, Vera outperforms leading open-source video editing models in content preservation while remaining competitive in edit quality, using 486K frames of layered training data.

We present Generative Relightable Avatars (GRA), a person-specific method for photorealistic free-view rendering and environment-map relighting of full-body humans. We postulate that modeling fine-grained appearance details is inherently a one-to-many problem that can benefit from a generative formulation. In contrast to fully regressive relightable avatar methods, GRA follows a hybrid approach that combines controllable, physics-grounded relighting with probabilistic refinement. Starting from a tracked animated mesh, we optimize material parameters in UV-space and render a coarse relit appearance under a target HDR environment map. Next, we refine the textures with a feed-forward model to capture pose-dependent texture dynamics and illumination effects beyond simplified reflectance assumptions. Finally, a fine-tuned video-to-video diffusion model transforms the physically grounded renderings into temporally coherent, high-detail videos while preserving 3D control, with an error-recycling strategy for generating long videos. Experimental evaluations demonstrate our method's improved perceptual quality over prior relightable avatar baselines. Project Page: https://vcai.mpi-inf.mpg.de/projects/GRA/

Diffusion-based video editing has made significant progress; however, achieving precise and temporally consistent object-level control, especially in multi-object scenarios, remains challenging due to attention leakage, identity drift, and unstable temporal dynamics. In this work, we propose IDAGEdit, a training-free framework for fine-grained multi-object video editing with strong temporal consistency. The framework adopts Layout-guided Attention Modulation to facilitate coherent multi-object editing, while Instance-level Masks are introduced to preserve individual object identity and enforce localized attention within each object region, thereby enabling fine-grained, object-level editing. Extensive qualitative and quantitative evaluations demonstrate that our method improves temporal stability and multi-object controllability over state-of-the-art video editing approaches.

Few-step distillation for video diffusion models has attracted significant attention, driven by the urgent demand for efficient deployment in real-world scenarios. However, Distribution Matching Distillation (DMD), a leading paradigm, tends to degrade under limited NFE budgets, manifesting in video generation as layout instability, oversaturation, and broken motion dynamics. We trace this failure to a structural limitation: standard DMD is an intra-sample distribution-matching objective with coordinate-wise gradients, and thus imposes no explicit constraint on the relational geometry across batch elements or temporal frames, leaving the underlying copula largely unregulated. Combined with the mode-seeking tendency of its reverse-KL objective, this absence of relational guidance makes DMD prone to collapsing into local optima in the few-step regime. Motivated by this insight, we propose Copula-aware DMD (CoDMD), a lightweight relational regularizer that reuses score estimates already produced by the frozen teacher and the online fake model to construct pairwise relation matrices across samples and frames. These are matched through a supplementary distributional objective that requires no additional networks, datasets, or sampling trajectories. On the Wan-2.1-T2V model series at 1.3B & 14B scales, CoDMD distills 50-step teachers into 4-step students, achieving an approximate 25$\times$ speed-up while attaining VBench scores of 84.46 & 84.87, outperforming prior trajectory-based (rCM 82.81 & 84.05) and distribution-based (DMD 83.38 & 83.81) methods.

Evaluating video generation with clean, pixel-based reward models disconnects evaluation from the noisy diffusion process and incurs massive VAE decoding costs. In this paper, we challenge this paradigm by asking a fundamental question: Can a powerful video generator inherently discriminate preferences directly from noisy latents? To answer this, we introduce \textbf{PRISM} (\textbf{P}reference \textbf{R}epresentation in \textbf{I}ntermediate \textbf{S}tates of Diffusion \textbf{M}odels). PRISM employs a lightweight Query-based Aggregation head with a frozen video diffusion backbone to decode preference signals from noisy latents. Surprisingly, PRISM not only achieves SOTA preference accuracy but also unlocks strong noise-robustness, which enables early-stage Best-of-$N$ sampling. This allows for filtering suboptimal candidates at the very beginning of denoising, drastically reducing computation while boosting video quality. We also reveal a strong positive correlation between a backbone's generative performance and its inherent evaluative power, enabling self-improving video backbones.

Diffusion models have achieved remarkable success in image and video generation and editing. While recent studies have extended these efforts toward motion editing, simultaneously transforming both motion and location-despite its practical importance-remains largely unexplored. To better understand robust motion-location editing, we first analyze the fundamental factors that degrade its quality. Based on this analysis, we propose TeleMorpher, one of the first one-shot frameworks to the best of our knowledge, for simultaneous motion-location editing. Our approach leverages motion priors, a target motion-centric video generated from an off-the-shelf model as motion-editing guidance, and the ground truth motion to enable more controllable and precise motion-location editing. Via this, our framework works as follows: (1) we first disentangle the protagonist and the background via pre-trained segmentation and inpainting models. (2) Then, we introduce a training-free pose warping that edits the protagonist's motion with the motion prior as the guidance. (3) The result of warped motion video is directly injected into a baseline motion editor during inference, mitigating the difference between source and target motions while preserving the appearance of the source video. (4) To enhance the reliability of quantitative evaluations, we propose two new LPIPS-based metrics that measure the background consistency before and after the motion editing and the fidelity of motion editing performance via measuring the difference between the extracted protagonist's skeletons from source and target videos. Experiments with in-the-wild videos and the TaiChi dataset demonstrate that TeleMorpher achieves superior performance across both quantitative and qualitative measurements (real-human evaluation), underscoring its effectiveness.

Cinematic compositing aims to integrate green-screen characters into novel environments while maintaining physical and photometric realism. Previous methods often fail to capture the complex bidirectional interactions between characters and their surroundings, which we characterize as Character-to-Environment (C2E) physical interaction and Environment-to-Character (E2C) lighting harmonization. To address this, we propose an end-to-end video diffusion framework that jointly models C2E and E2C interactions, specifically handling the challenges of interactive props. Our approach introduces a tri-mask-guided architecture with RGB-D joint denoising to ensure physically consistent interactions among the character, props, and environment. We further develop an efficient prior-driven data curation pipeline to construct high-quality relighting pairs without expensive rendering. Finally, a reference-conditioned mechanism enables controllable environment synthesis and precise prop replacement. Extensive experiments demonstrate that our framework significantly outperforms existing methods in cinematic-quality dynamic video compositing.

Recent progress has shown promise in distilling multi-step video diffusion models into efficient few-step students. Among them, Distribution Matching Distillation (DMD) and its successor DMD2 achieved strong generation quality and fast convergence. However, due to the nature of the reverse Kullback--Leibler (KL) objective, these methods exhibit two persistent failure modes: a substantial drop in sample diversity, and visibly over-saturated outputs that deviate from real-video appearance. In this work, we propose Data-Forcing Distillation (DFD), a simple post-training framework that restores diversity and fidelity in DMD with only a single-line of code change. At its core is the teacher score discrepancy to guide the student toward the real-data distribution, pulling it to missing modes (mitigating mode collapse) and away from problematic modes absent in real data (avoiding over-saturation). We provide an in-depth theoretical analysis of our framework and validate our approach on text-to-video, image-to-video, and autoregressive video generation. With only 100--300 steps of finetuning, DFD effectively restores diversity and fidelity on both Wan2.1-1.3B and Cosmos-Predict2.5-2B model, resolving the over-saturation artifacts with significantly better video dynamics and appearance, and even outperforms the teacher model.

Audio-visual speech enhancement (AVSE) exploits visual cues such as lip movements to recover speech in noisy environments. Recent work introduced diffusion-based unsupervised AVSE, where a speech diffusion model conditioned on visual features via cross-attention is trained and used as a data-driven prior for posterior sampling-based speech enhancement. Despite promising performance over its audio-only counterpart, the impact of explicitly enforcing cross-modal alignment in the fusion remains unclear. In this work, we propose to augment the diffusion training objective with a contrastive audio-visual loss to encourage stronger use of visual information while keeping the posterior sampling framework unchanged. Experiments across matched and mismatched test data show consistent improvements in interference suppression, signal reconstruction, and perceptual quality, with the largest gains at low SNRs. Code is available at https://github.com/ cexauce/AV-CA-DiffUSE

Open-weight video diffusion models can generate photorealistic unsafe content, from violence to misinformation, yet existing defenses either require expensive safety fine-tuning that degrades general capability, or apply external filters that are trivially bypassed by adversarial prompts. We present REINS (REpresentation-space INference-time Safety steering), a training-free method that aligns video diffusion models at inference time by steering their internal representations toward safe generation. Our key finding is that safety-relevant structure is linearly encoded in the hidden-state activations of video diffusion transformers, and a single direction, discovered via Supervised PCA on binary safety labels, suffices to separate safe from unsafe generation trajectories. At inference, adding this direction to hidden states at an intermediate transformer layer redirects generation from harmful content to semantically related safe alternatives, with no weight updates, no concept enumeration, and negligible computational overhead. Through mechanistic analysis, we reveal that while safety information accumulates monotonically with transformer depth, steering effectiveness peaks at intermediate layers (~50% depth), exposing a fundamental tradeoff between information availability and downstream propagation capacity. We evaluate REINS across 9 video diffusion models, multiple parameter scales (1.3B-5B), and both text-to-video and image-to-video generation, to our knowledge, the broadest safety evaluation suite in the video generation literature.

Large-scale video diffusion models often fail to preserve 3D structure over time, causing geometric drift and implausible motion under viewpoint changes. Existing methods usually enforce geometric consistency by using explicit geometry reconstructions, such as depth maps, point clouds, or reconstructed 3D structures, to define conditions, supervision, or reward signals, making the generator sensitive to errors from upstream geometry pipelines. We propose VideoWeave, a latent-space post-training framework that uses implicit geometry-model features to constrain the generative distribution, providing a more flexible and non-rigid form of guidance that mitigates the impact of reconstruction errors from geometry models. Specifically, VideoWeave adapts these features into geometry latents and jointly models them with video latents in a shared denoising space, allowing geometry to shape the generative distribution during training. To support this process, we build GeoVid-80K, an 80K-video dataset with paired appearance and geometry representations. Experiments on text-to-video and image-to-video generation show that VideoWeave improves geometric coherence while preserving strong visual quality. VideoWeave project page at https://videoweave.github.io/

While personalizing Image-to-Video (I2V) diffusion models with specific visual effects is increasingly demanded for high-end generation, current practice requires training a separate Low-Rank Adaptation (LoRA) module for each effect, incurring substantial data curation and iterative optimization costs that hinder interactive control. We present Prompt2Effect, a weight-driven hypernetwork that amortizes per-effect training by directly synthesizing effect-specific LoRA weights in a single forward pass. Unlike prior hypernetworks that regress adapter weights purely from semantics, Prompt2Effect is explicitly conditioned on the frozen base model weights, grounding prediction in the structural geometry of each layer. Furthermore, instead of predicting raw LoRA matrices, we introduce an SVD-canonicalized parameterization that resolves factorization ambiguity and stabilizes large-scale synthesis. Extensive experiments demonstrate that Prompt2Effect achieves on-par or superior video quality and effect alignment compared to conventional LoRA fine-tuning, while reducing the computational cost from 56 GPU training hours to 3.3 seconds of hypernetwork inference. When used as initialization for subsequent fine-tuning, our predicted weights further improve final performance and accelerate optimization by approximately 10x.

Cinematic video depicts multiple subjects acting or interacting at specific moments, captured with deliberate camera movement, and stitched together by shot transitions. Together, these elements demand a level of fine-grained control beyond current text-to-video models. Existing work addresses each axis in isolation: multi-subject personalization, temporal control, multi-shot synthesis, or camera control; no prior framework jointly integrates all four. We present CineOrchestra, a unified video diffusion model that controls subjects, events, cameras, and shot transitions simultaneously. Our key insight is that these heterogeneous cinematic elements share a fundamental structure: each is an entity acting over a specific temporal interval, which can therefore all be expressed through one shared structure of entity-centric conditioning primitives, augmented with reference images for visual entities. This formulation reduces the architectural challenge to a single positional encoding problem, which we solve with two parameter-free coordinated rotary embeddings: (a) an interval-sampled temporal RoPE that yields consistent attention behavior across events of dramatically varying duration, and (b) a 2D entity-temporal cross-attention RoPE that disambiguates per-entity conditions and routes each to its corresponding spatiotemporal region. On two new benchmarks, CineOrchestra outperforms six per-axis specialists on dense caption following and shot-transition timing, with consistent gains in a pairwise user study and component ablations. Project page: https://snap-research.github.io/CineOrchestra

We present Flex4DHuman, a multi-view video diffusion model that transforms a monocular or sparse multi-view video of a dynamic subject into synchronized dense multi-view videos using only relative camera-pose conditioning. Unlike prior human-centric methods that rely on skeletons, depth maps, normals, or rendered target-view geometry, Flex4DHuman requires no explicit geometry priors and instead conditions generation through relative camera-pose positional encoding. The generated videos can be directly ingested by downstream reconstruction pipelines to create dynamic 4D Gaussian splats. Built on the Wan 2.1 1.3B text-to-video model, Flex4DHuman preserves the backbone architecture and encodes camera and view information through a five-axis positional encoding that extends spatio-temporal RoPE with view indices and continuous SE(3) relative camera geometry. A three-stage curriculum progressively trains the model for pose following, flexible reference-to-target view generation, and temporal rollout. To support temporal rollout, we train with clean historical target-view tokens. We also add multi-view captions to enable test-time text control. Combined with an off-the-shelf 4D Gaussian Splatting stage, our framework lifts monocular static-camera videos into dynamic 4D Gaussian splats. Experiments on DNA-Rendering and ActorsHQ show that Flex4DHuman surpasses prior state-of-the-art methods, while the same formulation generalizes to animal categories after mixed human-animal training. These capabilities make Flex4DHuman a practical step toward scalable 4D content creation from casual monocular videos for simulation, gaming, AR/VR, and video re-shooting.

Event-based vision has drawn increasing attention owing to its distinctive properties, including ultra-high temporal resolution and extreme dynamic range. Recent works have introduced it to video super-resolution (VSR) to enhance flow estimation and temporal alignment. In contrast, this paper shifts the focus of event signals from motion refinement to texture enhancement in VSR. We propose EvTexture++, the first event-driven framework dedicated to texture enhancement in VSR. It leverages high-frequency spatiotemporal details from events to improve texture recovery. EvTexture++ incorporates a customized texture enhancement branch, along with an iterative texture enhancement module that progressively exploits high-temporal-resolution event information for texture restoration. This enables gradual refinement of texture regions across iterations, yielding more accurate and detailed high-resolution outputs. Besides intra-frame texture recovery, large motions could degrade inter-frame temporal consistency, particularly in texture regions, leading to texture flickering. To mitigate this, we further exploit the continuous-time motion cues of events to enhance temporal consistency, introducing a temporal texture alignment module that estimates event-guided texture-aware flow for precise inter-frame texture alignment. Moreover, EvTexture++ is designed as a plug-and-play tool to flexibly boost the performance of existing VSR models. Experiments on five datasets demonstrate that EvTexture++ achieves state-of-the-art performance. When integrated into recent VSR models, it yields significant improvements, with gains of up to 1.55 dB in PSNR on the texture-rich Vid4 dataset. Code: https://github.com/DachunKai/EvTexture.

Holistic visual tokenizers are fundamental to unified multimodal models (UMMs) as they map diverse visual inputs into a unified representation space. In this paper, we present HYDRA-X, the first UMM that unifies image and video tokenization within a single Vision Transformer (ViT). Our design is driven by two core challenges: efficiently injecting spatiotemporal reconstruction capability into a native ViT, and embedding image- and video-level semantic awareness into the latent space. To address the first, comprehensive ablations reveal two key findings: (1) frame-level causal temporal attention suffices for visual reconstruction, whereas full spatiotemporal attention degrades it; and (2) hierarchical temporal compression substantially outperforms single-step alternatives. To tackle the second, we propose a lightweight decompressor that upsamples temporally compressed features under joint image-video teacher supervision, thereby enforcing complementary semantic structures within the compact latent space. Building on this holistic tokenizer, we further propose a principled improvement of the editing pipeline: source-target interaction should occur at the latent level inside the tokenizer rather than at the semantic level inside the LLM, substantially improving editing consistency and accelerating convergence. Instantiated at the 7B dense model, HYDRA-X achieves strong performance across image and video understanding and generation tasks, paving the way for future unified-tokenizer UMMs.

Diffusion-based lip synchronization models achieve strong visual quality and audio-visual alignment, but full-sequence bidirectional attention and many denoising steps make them impractical for real-time inference. We present Lip Forcing, to our knowledge the first autoregressive diffusion method for video-to-video (V2V) lip synchronization, which distills a 14B audio-conditioned bidirectional video diffusion teacher into causal students. At inference, the students generate each chunk in only two denoising steps without inference-time CFG, enabling real-time lip synchronization. A lip-sync-specific teacher-trajectory analysis reveals a CFG fidelity-sync tradeoff: no-CFG predictions favor reference fidelity, whereas CFG-guided predictions favor synchronization within a mid-trajectory band. Lip Forcing translates this finding into three analysis-derived components: Sync-Window DMD, a two-step inference schedule, and a SyncNet-based reward. We validate Lip Forcing at two student scales, both distilled from the 14B teacher. The 1.3B student crosses into real-time streaming at 31 FPS, $17.6\times$ faster than its same-scale bidirectional model. The 14B student, the largest diffusion model reported for V2V lip synchronization, runs $39.8\times$ faster than its teacher at comparable reference fidelity. Time-to-first-frame is sub-millisecond at both scales, far below every diffusion baseline.

Modern Diffusion Transformers for video generation provide limited control over the progression of time and the editing of temporal dynamics. We propose a temporal-control methodology that extends a pretrained DiT with explicit time editing, allowing control over motion speed and temporal structure without redesigning the backbone. Its core implementation augments the pretrained model with a lightweight temporal module, preserving the original generative prior while expanding its controllable dynamic range.

Transitioning bidirectional video diffusion models into an autoregressive paradigm improves the interactivity of video world models, but existing causal pipelines need many stages (control fine-tuning, autoregressive training, causal initialization, few-step distillation) and still trail bidirectional models in quality due to error accumulation. Recent world models such as Yume-1.5 and Matrix-Game-3.0 instead adopt a bidirectional autoregressive approach, gaining fidelity and stable long-horizon rollout from self-correcting error propagation, yet open-source frameworks (e.g., minWM) support only causal models. We present BiWM, the first full-stack framework for interactive video world models under the bidirectional autoregressive paradigm, jointly optimizing generation quality and inference speed. From a pretrained video backbone, BiWM injects camera control by fine-tuning, then runs a few-step Distribution Matching Distillation (DMD) stage that turns the backbone into an action/camera-controllable world model: just two training stages instead of four in minWM, converging in a few hundred steps on 8xH200 GPUs. A single recipe spans Wan2.1-1.3B, Wan2.2-5B, HunyuanVideo-1.5-8B, and LTX-2.3-22B, and also supports secondary fine-tuning of existing bidirectional models. BiWM enables real-world camera control where minWM loses controllability, integrates pluggable history compression (FramePack-style and PackForcing-style) for long rollouts, and offers an optional NVFP4 4-bit training/inference pipeline. To counter DMD's mode-seeking degradation, we add GAN and mass-covering forward-KL objectives that preserve scene dynamics. We open-source BiWM for resource-constrained research and high-fidelity environment simulation.

Adapting large-scale pre-trained video generators for Video Super-Resolution (VSR) in novel domains remains computationally prohibitive. Methods that reformulate generation as direct Low-Quality to High-Quality mappings deviate from the original generative formulation, demanding extensive fine-tuning. ControlNet-style adapters lose their efficiency under modern Diffusion Transformers since the absence of encoder-decoder hierarchy forces duplication of the entire backbone. We observe that flow matching offers a principled alternative for cross-domain VSR adaptation. By predicting a constant velocity field across all timesteps, the adaptation task reduces to learning a fixed injection pattern rather than time-varying transformations. Building on this insight, we propose LiteVSR, a minimalist framework that performs VSR using a completely frozen Diffusion Transformer with a lightweight State-Aware Adapter. The adapter employs a dual-stream architecture that extracts static structural cues from the LQ input and dynamic cues from intermediate denoising states, aligning them through time-dependent cross-attention to enable adaptive transition from structural alignment to texture refinement as denoising proceeds. LiteVSR achieves competitive restoration quality with only 11.25% trainable parameters and 12 GPU-hours of training on a single A100, while maintaining fast sampling (down to a single step) compatibility.

Recent advances in Diffusion Transformers have driven rapid progress in video generation and editing, yet these capabilities are still handled by separate, task-specific models. Building a unified framework that supports diverse video tasks remains an open challenge: existing unified attempts either require dedicated auxiliary encoders or lack explicit mechanisms to distinguish heterogeneous conditioning tokens, struggling when the number and type of visual conditions vary across tasks. We propose TIDE, a unified framework that integrates instruction-based editing, reference-guided editing, and multi-reference generation. At its core, we introduce per-token task embeddings that assign each input token a task-specific identifier, enabling the model to explicitly disambiguate target, source, and reference tokens. To simultaneously capture high-level semantic understanding and fine-grained structural fidelity, we design a dual-path conditioning scheme that couples a vision-language model with a VAE latent path for complementary signals. We further devise a multi-task progressive training strategy that incrementally introduces tasks of increasing complexity, effectively harmonizing diverse objectives and enabling smooth generalization across heterogeneous task distributions. Extensive experiments on multiple video editing and generation benchmarks demonstrate that TIDE achieves state-of-the-art performance across all evaluated tasks. Our project page is available at https://LittleWork123.github.io/tide.

Human Image Animation has seen significant advancements, primarily driven by diffusion models. However, existing methods typically demand substantial training data and resources to achieve high-quality results, limiting generalization and accessibility. In this work, we introduce \emph{FreeAnimate}, a training-free framework that leverages the inherent capabilities of image diffusion models to enable temporal consistency, identity preservation, and background stability. Our approach incorporates a novel preview generation strategy that provides temporal and structural priors from generated preview frames, effectively guiding pose alignment and background consistency without training. Additionally, FreeAnimate introduces Inversion-Boosted Attention and Reference-Anchored Self-Attention modules to guarantee temporal consistency and identity preservation. Experimental results demonstrate that FreeAnimate outperforms existing training-free competitors and training-based baseline methods, achieving generation quality comparable to state-of-the-art methods and offering robust generalization across diverse datasets. Our project page is at https://freeani.github.io/.

Estimating hand-surface contact pressure from an egocentric view is crucial for AR/VR devices, robotic imitation, and ergonomic analysis. Existing methods often discretize pressure signal and process frames independently, leading to quantization errors and temporal inconsistencies. We present \emph{EgoPressDiff}, a conditional video diffusion framework that generates UV-pressure maps from visual input. The core of our approach is a multi-modal conditioning strategy, introducing a PoseNet and a Vertex Encoder to efficiently extract features from hand pose and 3D mesh vertices. These signals, along with depth information, guide the generative process to ensure the pressure fields are physically grounded. To effectively fuse these heterogeneous features, we further propose a Distribution-Calibrated Spatial Layer, which aligns their statistical properties before combination. Evaluated on the EgoPressure ego-view setting, EgoPressDiff achieves state-of-the-art results, improving Volumetric IoU by over 34% relative to prior baseline, while reducing MAE and maintaining high temporal accuracy. Our project page is at https://egopressdiff.github.io/.

The new era has witnessed a remarkable capability to extend Vision-Language Models (VLMs) for tackling tasks of video understanding. While current VLMs excel at event- or story-level understanding, their ability to capture fine-grained motion details remains limited, primarily due to their focus on high-level static semantic structures and macro-event logic. In contrast, Video Diffusion Models (VDMs) are adept at modeling dynamic motion patterns, benefiting from large-scale video data and the intrinsic requirement of temporal generation. In this paper, we introduce MotionEnhancer, a novel approach that leverages motion priors distilled from a powerful video diffusion model as auxiliary supervision to enhance the motion understanding capability of a VLM via attention alignment. MotionEnhancer comprises two simple parameter-free modules, Motion-sensitive Head Selection (MHS) and Motion-salient Text Token Identification (MTTI), to directly extract and optimize motion-related attentions from the VDM in a computation-only manner. MotionEnhancer provides a scalable solution for motion understanding without additional training parameters, modifications to existing architectures, or tool calling. Extensive experiments demonstrate that MotionEnhancer can achieve consistent improvements over state-of-the-art VLMs on two motion-level video understanding benchmarks, especially on motion-related metrics.

Image-to-Video diffusion models leverage input images to generate visually stunning content, yet frequently produce motion that violates physical laws. We reveal a surprising finding: a 2-step generation often exhibits better physical consistency than a 50-step output from the same model. Through spectral analysis, we trace this to phase erosion during denoising; the phase degrades significantly (dropping by $\approx 18\%$ from step 2 to step 50), whereas the magnitude remains relatively stable. Building on this insight, we propose PhaseLock, a training-free framework that preserves the valid motion priors from few-step inference throughout the denoising trajectory. Rather than relying on full-step inference for physical consistency, PhaseLock extracts a motion prior from just 2 steps and enforces it onto high-fidelity generation via Latent Delta Guidance. Our approach effectively mitigates phase degradation, improving physical consistency by an average of 6.2 points across diverse models while largely maintaining visual fidelity, with negligible overhead ($1.06\times$ time, $1.02\times$ memory) and reduced reliance on expensive external guidance methods ($\sim5\times$ time). Project Page: https://dnwjddl.github.io/phaselock

Video generation models based on Diffusion Transformers (DiTs) have achieved remarkable performance in video synthesis, yet they suffer from high inference latency and computational costs due to the quadratic complexity of 3D attention. Existing acceleration methods primarily reduce computational complexity within each individual denoising steps through techniques such as sparse attention and KV-caching. However, they rigidly adhere to the inherent constraint of the standard diffusion pipeline: every frame in the target video sequence must be subjected to a complete, dense denoising process across all diffusion timesteps. We observe that due to the corresponding contents and motions among adjacent frames, when keyframes with critical semantic transitions are anchored, the intermediate states of others often follow more predictable trajectories, which indicates that such uniform, dense denoising process is inherently redundant for natural video data. To this end, we introduce \textbf{RhymeFlow}, a training-free framework that decouples the denoising trajectories of different frames. Specifically, we first identify a sparse set of pivotal key frames that dominate the latent semantic evolution. Then, only these keyframes undergo dense, step-by-step denoising to ensure structural integrity, while non-keyframes progressively skip denoising steps to minimize computational cost. Since skipped intermediate states of non-keyframes break the temporal coherence in keyframe denoising steps, leading to visual degradation, we further introduce a latent trajectory projection module, which enables keyframes to interact with a complete and temporally consistent sequence representation. Extensive experiments on current DiT-based video generation models demonstrate our method outperforms existing baselines with higher inference speed and better visual quality.

Modern video diffusion models generate increasingly realistic and temporally coherent videos, motivating their use as candidate world simulators. Yet it remains unclear whether these models internally encode physical structure, or merely reproduce motion patterns seen during training. We study this question by probing video diffusion models along latent trajectories corresponding to real videos with known physical plausibility. To obtain such trajectories, we approximately invert the deterministic sampling process by integrating the learned velocity field backward from a clean video latent to noise, giving access to the model's intermediate states and attention maps. Using these recovered trajectories, we show that physical plausibility is linearly decodable from diffusion transformer states across IntPhys and InfLevel, reaching around 81.27% average accuracy and outperforming dedicated representation-learning baselines such as V-JEPA and VideoMAE. Surprisingly, this signal is absent from the VAE latent input and emerges inside the denoising transformer itself, despite the model not being trained with a self-supervised predictive objective. These findings suggest that physically meaningful representations can arise as a byproduct of generative denoising.

Video-Action Models (VAMs) leverage the broad visual dynamics captured by pre-trained video diffusion models, offering a promising path toward generalizable robot manipulation. However, RGB-only video rollouts are not directly actionable: they leave metric 3D motion, contact geometry, and fine-grained spatial constraints under-specified, making action grounding ambiguous. Meanwhile, scaling action supervision across diverse tasks and embodiments remains costly. We present PointAction, a framework that bridges video predictions to robot actions through explicit point-based 4D modeling. PointAction fine-tunes a foundation video generation model to jointly predict future RGB frames and dynamic 3D pointmaps, producing temporally consistent 3D motion of task-relevant scene geometry. These point dynamics serve as a structured, embodiment-agnostic action interface, which a diffusion-based action decoder maps to executable robot actions. By using metric 3D point dynamics as the interface between video prediction and control, PointAction reduces the ambiguity of RGB-only action grounding and supports transfer across tasks and embodiments with limited action supervision. Experiments show that PointAction achieves state-of-the-art 4D generation quality on robot scenes, outperforms existing baselines in simulation, and generalizes to two real robot arms unseen during pretraining.

Recent world action models leverage video foundation models by aligning broad visual-dynamics priors with executable robot actions. We revisit this alignment from a distributional perspective. Existing formulations typically narrow the aligned prior into an observation-conditioned policy distribution over future actions. In contrast, we keep the distribution broader by modeling the joint space of interaction videos and executable hand trajectories under multiple conditioning regimes. We propose Donk, a unified video-action denoising model for dexterous hands. With language, an initial image, and the initial hand state, Donk samples future videos and bimanual MANO trajectories as an action policy. Without the image condition, the same denoising architecture samples paired video-action rollouts from a text-conditioned distribution, turning the aligned video prior into a data engine. Across action, video, and text-only generation evaluations, Donk improves dexterous trajectory accuracy, preserves strong video fidelity, and produces smooth text-conditioned action rollouts under the same unified training recipe.

We present a diffusion-based method for relighting dynamic portrait videos with photorealism and temporal consistency. Our method is fueled by a hybrid training dataset that consists of real-captured and rendered dynamic portrait videos with diverse subject appearances, facial motions, head poses, and known lighting conditions. Specifically, we construct an LED-based lighting system for realistic lighting emulation and high-speed video relighting data acquisition. By leveraging the image priors embedded in pre-trained video diffusion models, and using per-frame high dynamic range (HDR) environment map as lighting control, we train a high-performance generative model for realistic and identity-preserving dynamic portrait video relighting. In addition to the environment map control, our model uses a synthesized background image to enable control on the camera's exposure level and color tone. Our model can produce temporally consistent relit portrait video that looks realistic and harmonious under a provided new environment and faithfully preserve the subject's expression and fine facial features, including skin tone, wrinkles, and facial hair. Our model generalizes well to unseen data, in terms of the subject appearance, motion, and lighting condition. We perform extensive experiments on relighting in-the-wild videos with various environment maps and demonstrate practical applications on portrait photography. Results show that our method achieves state-of-the-art performance in photorealism, lighting harmony, and temporal consistency.

Diffusion models have shown remarkable success in video generation. However, whether such models are truly aware of the 3D structure underlying visual observations, rather than simply reproducing plausible 2D projections, remains an open question. In this work, we investigate this question through human motion control, a task that requires precise modelling of 3D human geometry, motion, camera viewpoint, and scene context. Unlike prior methods that rely on rendered 2D motion guidance videos, we propose a render-free framework that conditions video generation directly on compressed 3D human mesh tokens. This representation preserves full 3D geometric information while enabling a unified token-based generation pipeline that processes video tokens jointly with motion tokens in a DiT-based architecture. This design requires the model to reason jointly about appearance, 3D structure, and camera viewpoint during video generation. Experimental results demonstrate strong performance on human motion control benchmarks, while reducing artifacts induced by view-dependent 2D guidance and trajectory-pose mismatches during editing. These findings suggest that video diffusion models, when equipped with mesh tokenization, can better capture complex 3D human structures and their interactions with the surrounding environment.

Modern vehicle platforms are equipped with a rich sensor suite, including LiDAR, calibrated multi-camera rigs, and accurate ego-motion, that in principle offers strong signal for re-rendering a driving scene from novel viewpoints. A growing line of recent work leverages video diffusion models for this task, using their generative priors to synthesize plausible novel views from sparse vehicle observations. In practice, however, existing methods exploit only a fragment of this signal, and their quality tends to degrade as the target trajectory departs from the recorded driving path. We argue that this is fundamentally a multi-sensor fusion problem: sparse LiDAR reprojections supply accurate but incomplete metric geometry, surround-view reference imagery supplies dense appearance but no metric depth, and camera poses tie the two together across views. We introduce StreetNVS, a video diffusion framework that jointly conditions on all three signals through a Reference-Enhanced Camera Attention module based on a relative ray-level positional encoding. We develop a two-stage curriculum training strategy that gradually exposes the model to increasingly sparse LiDAR. On the Waymo Open Dataset, StreetNVS substantially outperforms state-of-the-art baselines under sparse LiDAR conditioning, matches methods that rely on 10-100 times denser point clouds. We further show capabilities of synthesizing coherent videos along extreme out-of-trajectory paths such as elevation, lane-shift, pullback, and rotation. Our website: https://streetnvs.github.io

Large video diffusion models achieve strong visual quality but remain expensive to deploy because each sample requires many denoising steps and a large resident parameter footprint. This paper studies a deployment-oriented compression pipeline for Wan2.2-T2V-A14B by combining few-step distribution-matching distillation with low-bit quantization. The pipeline follows the model's dual-expert denoising route, calibrates the high-noise and low-noise branches separately, protects sensitive entrance layers, and uses HiF4-style low-bit representation to improve dynamic-range coverage. Quantization is calibrated on the distilled few-step student rather than on the original long-step trajectory, reducing activation-distribution mismatch during inference. The proposed co-design keeps the quantized model close to the same-step full-precision model and surpasses the original full-precision baseline at 8 and 20 steps on average. The 20-step setting gives the best quality-efficiency trade-off in the tested configurations.

While Video Diffusion Models (VDMs) excel at synthesizing high-fidelity videos, enabling precise camera and scene control remains challenging. Existing methods predominantly rely on implicit diffusion priors to generate unobserved regions, inevitably leading to structural collapse during high-dynamic movements or complex occlusions. To address this challenge, we propose Real2SAM2Real, a framework that leverages 3D lifting models (e.g., SAM3D) to extract an explicitly editable 3D cache, serving as a robust geometric scaffold for the VDM. By capturing the entire 3D volume of foreground entities rather than just their visible shells, this cache injects holistic spatial priors into the VDM, providing dependable 3D-aware guidance for complex scene dynamics. To effectively leverage this 3D guidance while preserving pre-trained priors, we design a Soft Spatial-Aligned Injection mechanism alongside a minimally invasive fine-tuning strategy tailored for VDMs. Furthermore, we employ masked normal maps as a cross-modal bridge to construct a 3D-free data curation and perturbation pipeline. Extensive experiments demonstrate that Real2SAM2Real enables precise, decoupled control over both camera trajectories and multi-entity motions. By utilizing the complementary context from generative 3D caches, our framework overcomes typical breakdowns caused by over-reliance on diffusion priors, maintaining exceptional spatiotemporal consistency under large camera shifts and severe occlusions. Crucially, by decoupling geometry from appearance, our VDM-tailored 3D cache eradicates perspective ambiguities caused by structural holes and erroneous facades, as well as misleading cues from reflections and refractions. Project website is available at https://jiayi-wu-leo.github.io/real2sam2real

Precise camera pose control is critical for video diffusion, yet maintaining geometric consistency remains a challenge. Existing methods that directly inject numerical camera parameters into the diffusion backbone often fail to bridge the gap between abstract coordinates and visual content, leading to structural distortions. To address this issue, we propose CameraNoise, a flow-to-noise warping method that encodes camera motion into a temporally coherent stochastic representation. Unlike conventional conditioning, CameraNoise embeds camera poses directly into the noise space. This decouples motion from scene appearance while faithfully preserving trajectory dynamics. Specifically, we introduce a novel Geometry-guided Reprojection Flow and a noise warping algorithm, which jointly preserve the Gaussian prior of diffusion and ensure consistent noise propagation under camera transformations. By integrating CameraNoise into the diffusion process, our framework delivers stable, high-fidelity videos. Extensive experiments demonstrate that our approach significantly outperforms prior methods in both visual quality and trajectory faithfulness. The project page and code are available at: https://gulucaptain.github.io/CameraNoise/.

Recent progress in video diffusion models has enabled remarkable generative fidelity, yet leveraging these priors for restoration remains limited by the strong coupling between conditional and unconditional branches in standard classifier-free guidance. We introduce a training-free framework that enhances distorted and low-resolution videos by decoupling these signals in time. Our proposed Decoupled Time Guidance (DTG) evaluates the unconditional branch at a cleaner diffusion timestep, providing a lookahead prior that preserves geometry while suppressing replication of warped content. This temporal bias is annealed throughout sampling, allowing the model to transition from structure correction to detail refinement without retraining. Combined with any off-the-shelf restoration module in a plug-and-play manner, our approach improves perceptual coherence and restores plausible structure in AIgenerated and real-world videos alike. To facilitate evaluation, we curate GenWarp480, a benchmark of 4,400 distorted 480p videos synthesized from diverse text-to-video models. GenWarp480 focuses on characteristic generative degradations such as warped faces, body misalignments, and spatial artifacts, providing a purpose-built testbed for assessing robustness to generative errors. Extensive experiments demonstrate that our method achieves significant improvements in structural fidelity and temporal stability without any model training.

Scaling Diffusion Transformers to generate high-resolution, long videos is constrained by the quadratic cost of self-attention, and existing sparse attention methods degrade under high sparsity. We show empirically that generation quality is determined not by the sparsity ratio itself, but by how well the sparse mask aligns with the tile-wise geometry of full attention. Based on this insight, we propose Veda, a distilled sparse attention framework that formulates tile selection as an explicit reconstruction problem from full attention. Veda integrates statistics-aware tile scoring with head-aware tiling to reduce estimation error and structural mismatch, enabling aggressive sparsity. A hardware-efficient tile-skipping kernel converts theoretical sparsity into practical wall-clock speedups. Experiments on large video diffusion models, including Waver and Wan2.1, demonstrate substantial acceleration with no noticeable degradation in generation quality. To generate 720P 10-second videos on Waver-T2V-12B, Veda achieves a 5.1$\times$ end-to-end speedup and a 10.5$\times$ self-attention speedup, reducing attention overhead from 92% to 50%. Notably, the gains increase with sequence length, indicating that Veda scales favorably with spatiotemporal resolution across models.

Digital humans are fundamental to immersive interaction, yet creating a unified model for holistic modalities, including text, audio, motion, and visual content, remains an open challenge. In this paper, we present Archon, a fully pretrained, human-centric unified multimodal model for holistic avatar generation. Archon unifies seven modalities with modality-specific tokenizers, and a native autoregressive unified multimodal model pretrained on synchronized modalities and 72 diverse tasks to model holistic joint distributions. To address the token explosion challenge in high-fidelity talking videos, we introduce a memory-efficient semantic video reparameterization, achieving 4x token reduction while preserving fine-grained dynamics, coupled with a semantic-driven video diffusion decoder. We further propose a "Thinking in Modality" that decomposes ambiguous cross-modal tasks into stepwise thinking in an alternative chain of modality, progressively enhancing fidelity and controllability. Extensive experiments demonstrate that Archon achieves superior or comparable performance across diverse digital human generation tasks, validating the effectiveness of our unified framework. Project page: https://zju3dv.github.io/archon/.

Distribution Matching Distillation (DMD) is a widely used paradigm for accelerating inference in few-step video diffusion models. However, DMD-style video distillation faces two coupled challenges: the fake score must track a continuously evolving generator, making training costly when frequent updates are required, while reverse-KL-style matching can be mode-seeking and conservative for preserving strong motion dynamics. To address these issues, we propose \textbf{Score Gradient Matching Distillation (SGMD)}. SGMD adopts a fake-score perspective by directly optimizing the fake score toward the teacher, while using teacher stop-gradient Fisher as a stable distribution-matching objective. We provide a gradient analysis that motivates this objective choice under ideal tracking. Building on this, SGMD introduces a pair of dual potentials: negative-residual (NR) for outer-loop correction and residual-contraction (RC) for inner-loop tracking. Empirically, compared to DMD2, SGMD achieves an approximately $\sim 3\times$ training speedup and substantially improves motion dynamics for 4-step distilled models while preserving temporal consistency. A human study confirms that SGMD is preferred in motion quality and overall preference, while visual quality and text alignment remain comparable. Code is available at https://github.com/ModelTC/LightX2V.

We present a method for harmonizing the lighting of a foreground video to match a target background scene, adjusting shadows, color tone, and illumination intensity (relightful harmonization). Unlike images, acquiring labeled data for videos, where identical motions are recorded under different lighting conditions, is practically infeasible and non-scalable. While one way to create such paired data is to apply existing image-based harmonization models frame by frame to a video, the resulting outputs often suffer from significant temporal jitters. We overcome this problem by introducing a novel lighting deflickering model that can stabilize the global and local lighting flickering artifacts. Our video diffusion model learns from these upgraded deflickered data with a volume of real and synthetic videos to generate high-quality video harmonization results. We further propose an asymmetric alpha mask conditioning technique to learn the clean boundaries from real videos. Experiments demonstrate that our model achieves strong temporal coherence, naturalness, cleaner boundaries, and physically meaningful lighting behavior, while maintaining strong relighting expressiveness compared to prior image-based and video-based harmonization methods.

Unsupervised visual object tracking is a challenging task that requires following arbitrary targets in videos without training on ground-truth annotations. Despite considerable progress, existing state-of-the-art unsupervised trackers often struggle in scenarios that demand fine-grained understanding of semantic and visual structural information within video frames. Text-to-image diffusion models are well known for their ability to generate images that accurately reflect the semantics and structures described in the input prompt, demonstrating a strong grasp of visual semantics and structures. Building on this capability, we approach the unsupervised tracking from a new perspective by exploiting the rich semantic knowledge encoded in pretrained text-to-image diffusion models. To adapt the diffusion models, which are originally developed for image generation, to the tracking task, we reinterpret the models as a bridge between text and image modalities. This connection is realized through the cross-attention mechanism: when both text and an image are input into the models, they highlight the regions of the image that are semantically aligned with the text in the cross-attention maps. We therefore learn a prompt that represents the tracking target and activates its corresponding region in the cross-attention map for each frame, which enables object tracking with the diffusion model. Specifically, our method Diff-Tracking is composed of two main components: an initial prompt learner and an online prompt updater. The initial prompt learner generates a prompt that captures the target object in the first frame, allowing the diffusion model to identify the target. The online prompt updater refines the prompt based on motion information, enabling consistent tracking across video frames. We evaluate our approach on six challenging tracking datasets demonstrate the effectiveness of our approach.

Controllable and physically grounded egocentric video generation is essential for embodied agents to reason about how their own and others' actions manifest and change the world. Compared to generic video synthesis, egocentric generation is especially challenging: the camera is tightly coupled to the actor, leading to rapid viewpoint changes and frequent self-occlusions; the underlying actions are subtle, articulated, and often only partially visible; and both the people and the scene state must evolve consistently with the specified controls. We present E$^3$C, a controllable video diffusion framework for egocentric generation that builds structured and compact conditions disentangling persistent scene structure from human-driven dynamics. From context frames, E$^3$C constructs a semi-dense point cloud-based 3D memory and augments each point with appearance descriptors from video-VAE features. Rendering this memory into target viewpoints produces conditioning aligned with the target frames. Human dynamics are modeled separately. The observed people in the scene are controlled by skeleton renderings (exo human control), while the camera wearer is specified by their 3D body joints and 6DoF wrist motion (ego human control). To preserve ego human control when the wearer's body parts are invisible, we introduce an ego motion encoder that produces persistent cross-attention tokens. Experiments on Nymeria show that E$^3$C improves visual fidelity, camera-motion accuracy, object consistency, and ego & exo human control over strong baselines, while also enabling intuitive scene editing.

Event cameras offer significant advantages over conventional frame-based counterparts, including high temporal resolution, low latency, and energy efficiency. These characteristics make them suitable for high-speed and high-dynamic range scene acquisition scenarios; however, the lack of dense intensity frames limits the direct applicability of conventional computer vision methods for scene understanding. Event-to-video (E2V) reconstruction seeks to bridge this gap by converting asynchronous event streams into a sequence of synchronous video frames. Existing E2V reconstruction methods based on convolutional neural networks and transformers operate primarily in the spatial domain and often struggle to recover fine structural details while suppressing severe reconstruction artifacts. To address these issues, we propose MSFET-E2V, a novel multiscale frequency-enhanced transformer model. At its core lies a cross-domain attention module, which fuses spatio-temporal features with frequency-aware representations derived from the discrete wavelet transform. Unlike prior methods relying solely on spatial attention, our approach effectively captures both local and global structures by taking into account low- and high-frequency components, enhancing detail preservation and robustness across various motion scenarios. Furthermore, we propose a lightweight wavelet-enhanced skip block that serves as a skip connection, facilitating artifact suppression and structural detail refinement through joint spatial-frequency domain processing. Extensive experiments demonstrate that MSFET-E2V achieves superior performance over state-of-the-art methods on multiple real-world event datasets, offering significant gains in reconstruction quality. Moreover, compared to the existing transformer-based method, our proposed model significantly reduces the number of parameters, the GPU memory usage, and inference time.

Consistent 3D geometry estimation from streaming RGB input is crucial for real-world applications such as autonomous driving, embodied AI, and large-scale reconstruction. While modern monocular geometry foundation models achieve strong single-image accuracy, they exhibit severe temporal inconsistency on continuous input, notably dominated by scale--shift drifting. Through targeted empirical analysis, we trace this instability to its root cause: fluctuations in latent feature statistics, whose mean and variance directly determine the predicted depth's scale and shift. Building on this insight, we introduce Dynamic Feature Normalization (DyFN), a lightweight, causal recurrent module that dynamically and robustly modulates feature statistics to maintain stable geometry over time. We adapt powerful pretrained monocular geometry models for streaming by finetuning only DyFN, a mere 2% additional parameters, while keeping the backbone frozen, thereby achieving temporal consistency without compromising single-image accuracy. Extensive experiments across four benchmarks show that DyFN effectively eliminates temporal artifacts such as disjointed layering and positional jitter, and achieves state-of-the-art temporal stability, improving over prior streaming methods by up to 14% and even outperforming heavier non-causal video baselines. Project Page: https://shawlyu.github.io/DyFN

Instruction-based video editing requires transforming a source video according to a natural-language instruction while preserving irrelevant content and remaining temporally coherent. We argue that existing Diffusion Transformer (DiT) editors struggle with this task for two structural reasons. First, conditioning signals are fed undifferentiated into all transformer blocks, forcing a single token stream to encode both global editing intent and fine-grained visual evidence. Second, the cross-attention patterns that govern the edit are supervised only indirectly through pixel-level reconstruction, leaving the model's internal reasoning process under-constrained. To address both limitations, we propose RVEDiT, an implicit Reasoning Video Editing DiT framework built around two complementary components. The first, Granularity-Routed Token Conditioning, introduces learnable editing tokens distilled from a multimodal LLM and routes them to shallow blocks, while reserving native visual and textual tokens for deeper blocks, thereby inducing a coarse-to-fine editing process inside the backbone. The second, Reference-Anchored Attention Alignment, employs a parameter-sharing reference branch during training and maximizes the mutual information between the attention features of the editing and reference branches, regularizing the model's internal reasoning without incurring any additional inference cost. Experiments on standard instruction-based video editing benchmarks show that RVEDiT consistently outperforms state-of-the-art baselines, with particularly strong gains on localized and compositional edits.

Recent progress of video diffusion models have enabled extensive simulation of the physical world. While simulation with hand object interaction has been less explored. We propose DexSIM, a dexterous simulation framework for simulating dexterous manipulation in real-time. While previous works utilizing video diffusion and 3D reconstruction focus on navigation, dexterous manipulation has been limited while it has extensive applications for creating interactive experiences with the simulated world and for generating synthetic data for robotics. Existing methods lack real-time interactivity and long-term spatial consistency and memory. We propose a 2-stage training framework for DexSIM. First we train a bi-directional video diffusion model by jointly embedding the hand action trajectory and video in a unified feature space. We utilize gaussian heatmap hand encoding for more accurate hand representation. Then we conduct a roll-out based autoregressive training with updated spatial cache as attention sink for spatial memory, which improves long-term consistency and 3D aware dexterous manipulation simulation. DexSIM outperforms the baseline on pixel and semantic similarity, motion fidelity, and hand projection accuracy. It also allows new applications such as hand motion transfer and runs at 15.24 FPS real-time interactivity.

Diffusion transformers have achieved remarkable success in high-quality video generation, yet their reliance on spatiotemporal 3D full attention incurs prohibitive computational cost due to the quadratic complexity of attention. Block sparse attention is a common approach to mitigate this by focusing computation on important regions. However, attention maps in DiTs exhibit inherently dynamic and fine-grained sparsity, which causes existing block sparse attention methods to degrade significantly in quality, especially at high sparsity ratios. In this paper, we revisit block sparse attention and derive a theoretical lower bound on attention recall to characterize the key factors governing its effectiveness. Guided by these insights, we propose DFSAttn, a training-free sparse attention framework that enables dynamic, fine-grained sparsification efficiently. DFSAttn incorporates three core designs: Hilbert curve-based token reordering to achieve fine-grained sparsity while preserving efficient GPU execution, hierarchical block scoring for accurate block importance estimation, and sparse mask caching with adaptive ratios to balance accuracy and efficiency. Experimental results demonstrate that DFSAttn consistently outperforms prior methods under high sparsity, achieving up to 2.1$\times$ end-to-end speedup while maintaining high generation quality. Our code is open-sourced and available at https://github.com/jessica-hujie/DFSAttn.

Diffusion Transformer (DiT) has emerged as a powerful model architecture for generating high-quality images and videos. In the case of video DiT, 3D Spatio-Temporal Attention increases token length in proportion to the number of frames, sharply increasing computational cost. Token reduction methods mitigate this cost by exploiting spatial redundancy, but existing approaches rely on inaccurate similarity estimates and lightweight matching algorithms, resulting in poor matching quality and only marginal acceleration. To overcome these limitations, we propose ORBIS, an SW-HW co-designed accelerator for video DiT. ORBIS leverages the output activation from the previous timestep to obtain more accurate inter-token similarity, substantially improving matching quality and enabling a higher token reduction ratio. We further introduce a Distribution-Aware Token Matching (DATM) algorithm that captures global token distribution and explicitly minimizes token-pair loss for additional gains. To fully hide DATM latency, we design specialized, deeply pipelined hardware and minimize its hardware cost through quantization, occupying only 2.4% of total area with negligible accuracy loss. Extensive experiments show that ORBIS achieves about 2x higher token reduction ratio than the state-of-the-art approach, AsymRnR, while delivering up to 4.5x speedup and 79.3% energy reduction compared to an NVIDIA A100 GPU.

Video Virtual Try-On (VVT) aims to seamlessly replace a garment on a person in a video with a new one. While existing methods have made significant strides in maintaining temporal consistency, they are predominantly confined to non-interactive scenarios where models merely showcase garments. This limitation overlooks a crucial aspect of real-world apparel presentation: active human-garment interaction. To bridge this gap, we introduce and formalize a new challenging task: Interactive Video Virtual Try-On (Interactive VVT), where subjects in the video actively engage with their clothing. This task introduces unique challenges beyond simple texture preservation, including: (1) resolving the semantic ambiguity of interactions from standard pose information, and (2) learning complex garment deformations from video where interactive moments are sparse and brief. To address these challenges, we propose iTryOn, a novel framework built upon a large-scale video diffusion Transformer. iTryOn pioneers a multi-level interaction injection mechanism to guide the generation of complex dynamics. At the spatial level, we introduce a garment-agnostic 3D hand prior to provide fine-grained guidance for precise hand-garment contact, effectively resolving spatial ambiguity. At the semantic level, iTryOn leverages global captions for overall context and time-stamped action captions for localized interactions, synchronized via our novel Action-aware Rotational Position Embedding (A-RoPE). Extensive experiments demonstrate that iTryOn not only achieves state-of-the-art performance on traditional VVT benchmarks but also establishes a commanding lead in the new interactive setting, marking a significant step towards more dynamic and controllable virtual try-on experiences.