CVPR 2025 Papers

Recent advancements in visual generation technologies have markedly increased the scale and availability of video datasets, which are crucial for training effective video generation models. However, a significant lack of high-quality, human-centric video datasets presents a challenge to progress in this field. To bridge this gap, we introduce OpenHumanVid, a large-scale and high-quality human-centric video dataset characterized by precise and detailed captions that encompass both human appearance and motion states, along with supplementary human motion conditions, including skeleton sequences and speech audio. To validate the efficacy of this dataset and the associated training strategies, we propose an extension of existing classical diffusion transformer architectures and conduct further pretraining of our models on the proposed dataset. Our findings yield two critical insights: First, the incorporation of a large-scale, high-quality dataset substantially enhances evaluation metrics for generated human videos while preserving performance in general video generation tasks. Second, the effective alignment of text with human appearance, human motion, and facial motion is essential for producing high-quality video outputs. Based on these insights and corresponding methodologies, the straightforward extended network trained on the proposed dataset demonstrates an obvious improvement in the generation of human-centric videos.

Transformers today still struggle to generate one-minute videos because self-attention layers are inefficient for long context. Alternatives such as Mamba layers struggle to produce coherent scenes because their hidden states are small and less expressive. We experiment with Test-Time Training (TTT) layers, whose hidden states themselves can be neural networks, therefore larger and more expressive. Adding TTT layers into a pre-trained Transformer enables it to generate one-minute videos from text storyboards. We curate a dataset based on Tom and Jerry cartoons as a proof-of-concept benchmark. Compared to baselines such as Mamba 2, Gated DeltaNet, and sliding-window attention layers, TTT layers generate much more coherent videos that tell complete stories, leading by 34 Elo points in a human evaluation of 100 videos per method. Although promising, our results are still limited in physical realism, and the efficiency of our implementation can be further improved. Sample videos, code and annotations are available at: https://test-time-training.github.io/video-dit

Video large language models (Video-LLMs) can temporally ground language queries and retrieve video moments. Yet, such temporal comprehension capabilities are neither well-studied nor understood. So we conduct a study on prediction consistency -- a key indicator for robustness and trustworthiness of temporal grounding. After the model identifies an initial moment within the video content, we apply a series of probes to check if the model's responses align with this initial grounding as an indicator of reliable comprehension. Our results reveal that current Video-LLMs are sensitive to variations in video contents, language queries, and task settings, unveiling severe deficiencies in maintaining consistency. We further explore common prompting and instruction-tuning methods as potential solutions, but find that their improvements are often unstable. To that end, we propose event temporal verification tuning that explicitly accounts for consistency, and demonstrate significant improvements for both grounding and consistency. Our data and code are open-sourced at https://github.com/minjoong507/Consistency-of-Video-LLM.

Video diffusion models have shown great potential in generating high-quality videos, making them an increasingly popular focus. However, their inherent iterative nature leads to substantial computational and time costs. Although techniques such as consistency distillation and adversarial training have been employed to accelerate video diffusion by reducing inference steps, these methods often simply transfer the generation approaches from Image diffusion models to video diffusion models. As a result, these methods frequently fall short in terms of both performance and training stability. In this work, we introduce a two-stage training framework that effectively combines consistency distillation with adversarial training to address these challenges. Additionally, we propose a novel video discriminator design, which eliminates the need for decoding the video latents and improves the final performance. Our model is capable of producing high-quality videos in merely one-step, with the flexibility to perform multi-step refinement for further performance enhancement. Our quantitative evaluation on the OpenVid-1M benchmark shows that our model significantly outperforms existing methods. Notably, our 1-step performance (FVD 171.15) exceeds the 8-step performance of the consistency distillation based method, AnimateLCM (FVD 184.79), and approaches the 25-step performance of advanced Stable Video Diffusion (FVD 156.94).

Current image generation models can effortlessly produce high-quality, highly realistic images, but this also increases the risk of misuse. In various Text-to-Image or Image-to-Image tasks, attackers can generate a series of images containing inappropriate content by simply editing the language modality input. To mitigate this security concern, numerous guarding or defensive strategies have been proposed, with a particular emphasis on safeguarding language modality. However, in practical applications, threats in the vision modality, particularly in tasks involving the editing of real-world images, present heightened security risks as they can easily infringe upon the rights of the image owner. Therefore, this paper employs a method named typographic attack to reveal that various image generation models are also susceptible to threats within the vision modality. Furthermore, we also evaluate the defense performance of various existing methods when facing threats in the vision modality and uncover their ineffectiveness. Finally, we propose the Vision Modal Threats in Image Generation Models (VMT-IGMs) dataset, which would serve as a baseline for evaluating the vision modality vulnerability of various image generation models.

Recent advancements in text-to-video models such as Sora, Gen-3, MovieGen, and CogVideoX are pushing the boundaries of synthetic video generation, with adoption seen in fields like robotics, autonomous driving, and entertainment. As these models become prevalent, various metrics and benchmarks have emerged to evaluate the quality of the generated videos. However, these metrics emphasize visual quality and smoothness, neglecting temporal fidelity and text-to-video alignment, which are crucial for safety-critical applications. To address this gap, we introduce NeuS-V, a novel synthetic video evaluation metric that rigorously assesses text-to-video alignment using neuro-symbolic formal verification techniques. Our approach first converts the prompt into a formally defined Temporal Logic (TL) specification and translates the generated video into an automaton representation. Then, it evaluates the text-to-video alignment by formally checking the video automaton against the TL specification. Furthermore, we present a dataset of temporally extended prompts to evaluate state-of-the-art video generation models against our benchmark. We find that NeuS-V demonstrates a higher correlation by over 5x with human evaluations when compared to existing metrics. Our evaluation further reveals that current video generation models perform poorly on these temporally complex prompts, highlighting the need for future work in improving text-to-video generation capabilities. We open-source our benchmark, code, and dataset at https://utaustin-swarmlab.github.io/neusv.

In recent text-video retrieval, the use of additional captions from vision-language models has shown promising effects on the performance. However, existing models using additional captions often have struggled to capture the rich semantics, including temporal changes, inherent in the video. In addition, incorrect information caused by generative models can lead to inaccurate retrieval. To address these issues, we propose a new framework, Narrating the Video (NarVid), which strategically leverages the comprehensive information available from frame-level captions, the narration. The proposed NarVid exploits narration in multiple ways: 1) feature enhancement through cross-modal interactions between narration and video, 2) query-aware adaptive filtering to suppress irrelevant or incorrect information, 3) dual-modal matching score by adding query-video similarity and query-narration similarity, and 4) hard-negative loss to learn discriminative features from multiple perspectives using the two similarities from different views. Experimental results demonstrate that NarVid achieves state-of-the-art performance on various benchmark datasets.

Efficiently retrieving and synthesizing information from large-scale multimodal collections has become a critical challenge. However, existing video retrieval datasets suffer from scope limitations, primarily focusing on matching descriptive but vague queries with small collections of professionally edited, English-centric videos. To address this gap, we introduce MultiVENT 2.0, a large-scale, multilingual event-centric video retrieval benchmark featuring a collection of more than 218,000 news videos and over 3,900 queries targeting specific world events. These queries specifically target information found in the visual content, audio, embedded text, and text metadata of the videos, requiring systems leverage all these sources to succeed at the task. Preliminary results show that state-of-the-art vision-language models struggle significantly with this task, and while alternative approaches show promise, they are still insufficient to adequately address this problem. These findings underscore the need for more robust multimodal retrieval systems, as effective video retrieval is a crucial step towards multimodal content understanding and generation.

Video personalization methods allow us to synthesize videos with specific concepts such as people, pets, and places. However, existing methods often focus on limited domains, require time-consuming optimization per subject, or support only a single subject. We present Video Alchemist--a video model with built-in multi-subject, open-set personalization capabilities for both foreground objects and background, eliminating the need for time-consuming test-time optimization. Our model is built on a new Diffusion Transformer module that fuses each conditional reference image and its corresponding subject-level text prompt with cross-attention layers. Developing such a large model presents two main challenges: dataset and evaluation. First, as paired datasets of reference images and videos are extremely hard to collect, we sample selected video frames as reference images and synthesize a clip of the target video. However, while models can easily denoise training videos given reference frames, they fail to generalize to new contexts. To mitigate this issue, we design a new automatic data construction pipeline with extensive image augmentations. Second, evaluating open-set video personalization is a challenge in itself. To address this, we introduce a personalization benchmark that focuses on accurate subject fidelity and supports diverse personalization scenarios. Finally, our extensive experiments show that our method significantly outperforms existing personalization methods in both quantitative and qualitative evaluations.

Recent advancements in video generation models, such as Stable Video Diffusion, have shown promising results, but these works primarily focus on short videos, often limited to a single scene and lacking a rich storyline. These models struggle with generating long videos that involve multiple scenes, coherent narratives, and consistent characters. Furthermore, there is currently no publicly accessible dataset specifically designed for analyzing, evaluating, and training models for long video generation. In this paper, we present MovieBench: A Hierarchical Movie-Level Dataset for Long Video Generation, which addresses these challenges by providing unique contributions: (1) character consistency across scenes, (2) long videos with rich and coherent storylines, and (3) multi-scene narratives. MovieBench features three distinct levels of annotation: the movie level, which provides a broad overview of the film; the scene level, offering a mid-level understanding of the narrative; and the shot level, which emphasizes specific moments with detailed descriptions.

Video personalization, which generates customized videos using reference images, has gained significant attention.However, prior methods typically focus on single-concept personalization, limiting broader applications that require multi-concept integration.Attempts to extend these models to multiple concepts often lead to identity blending, which results in composite characters with fused attributes from multiple sources.This challenge arises due to the lack of a mechanism to link each concept with its specific reference image.We address this with anchored prompts, which embed image anchors as unique tokens within text prompts, guiding accurate referencing during generation.Additionally, we introduce concept embeddings to encode the order of reference images.Our approach, Movie Weaver, seamlessly weaves multiple concepts--including face, body, and animal images--into one video, allowing flexible combinations in a single model.The evaluation shows that Movie Weaver outperforms existing methods for multi-concept video personalization in identity preservation and overall quality.

The image-to-video (I2V) generation is conditioned on the static image, which has been enhanced recently by the motion intensity as an additional control signal. These motion-aware models are appealing to generate diverse motion patterns, yet there lacks a reliable motion estimator for training such models on large-scale video set in the wild. Traditional metrics, e.g., SSIM or optical flow, are hard to generalize to arbitrary videos, while, it is very tough for human annotators to label the abstract motion intensity neither. Furthermore, the motion intensity shall reveal both local object motion and global camera movement, which has not been studied before. This paper addresses the challenge with a new motion estimator, capable of measuring the decoupled motion intensities of objects and cameras in video. We leverage the contrastive learning on randomly paired videos and distinguish the video with greater motion intensity. Such a paradigm is friendly for annotation and easy to scale up to achieve stable performance on motion estimation. We then present a new I2V model, named MotionStone, developed with the decoupled motion estimator. Experimental results demonstrate the stability of the proposed motion estimator and the state-of-the-art performance of MotionStone on I2V generation. These advantages warrant the decoupled motion estimator to serve as a general plug-in enhancer for both data processing and video generation training.

Animating images with interactive motion control has garnered popularity for image-to-video (I2V) generation. Modern approaches typically rely on large Gaussian kernels to extend motion trajectories as condition without explicitly defining movement region, leading to coarse motion control and failing to disentangle object and camera moving. To alleviate these, we present MotionPro, a precise motion controller that novelly leverages region-wise trajectory and motion mask to regulate fine-grained motion synthesis and identify target motion category (i.e., object or camera moving), respectively. Technically, MotionPro first estimates the flow maps on each training video via a tracking model, and then samples the region-wise trajectories to simulate inference scenario. Instead of extending flow through large Gaussian kernels, our region-wise trajectory approach enables more precise control by directly utilizing trajectories within local regions, thereby effectively characterizing fine-grained movements. A motion mask is simultaneously derived from the predicted flow maps to capture the holistic motion dynamics of the movement regions. To pursue natural motion control, MotionPro further strengthens video denoising by incorporating both region-wise trajectories and motion mask through feature modulation. More remarkably, we meticulously construct a benchmark, i.e., MC-Bench, with 1.1K user-annotated image-trajectory pairs, for the evaluation of both fine-grained and object-level I2V motion control. Extensive experiments conducted on WebVid-10M and MC-Bench demonstrate the effectiveness of MotionPro. Please refer to our project page for more results: https://zhw-zhang.github.io/MotionPro-page/.

Motion control is crucial for generating expressive and compelling video content; however, most existing video generation models rely mainly on text prompts for control, which struggle to capture the nuances of dynamic actions and temporal compositions. To this end, we train a video generation model conditioned on spatio-temporally sparse or dense motion trajectories. In contrast to prior motion conditioning work, this flexible representation can encode any number of trajectories, object-specific or global scene motion, and temporally sparse motion; due to its flexibility we refer to this conditioning as motion prompts. While users may directly specify sparse trajectories, we also show how to translate high-level user requests into detailed, semi-dense motion prompts, a process we term motion prompt expansion. We demonstrate the versatility of our approach through various applications, including camera and object motion control, "interacting" with an image, motion transfer, and image editing. Our results showcase emergent behaviors, such as realistic physics, suggesting the potential of motion prompts for probing video models and interacting with future generative world models. Finally, we evaluate quantitatively, conduct a human study, and demonstrate strong performance.

Text-Image-to-Video (TI2V) generation aims to generate a video from an image following a text description, which is also referred to as text-guided image animation. Most existing methods struggle to generate videos that align well with the text prompts, particularly when motion is specified. To overcome this limitation, we introduce MotiF, a simple yet effective approach that directs the model's learning to the regions with more motion, thereby improving the text alignment and motion generation. We use optical flow to generate a motion heatmap and weight the loss according to the intensity of the motion. This modified objective leads to noticeable improvements and complements existing methods that utilize motion priors as model inputs. Additionally, due to the lack of a diverse benchmark for evaluating TI2V generation, we propose TI2V-Bench, a dataset consists of 320 image-text pairs for robust evaluation. We present a human evaluation protocol that asks the annotators to select an overall preference between two videos followed by their justifications. Through a comprehensive evaluation, MotiF outperforms nine open-sourced models, achieving an average preference of 72%. The TI2V-Bench and additional results are released in https://wang-sj16.github.io/motif/.

Visual-Language Tracking (VLT) is emerging as a promising paradigm to bridge the human-machine performance gap. For single objects, VLT broadens the problem scope to text-driven video comprehension. Yet, this direction is still confined to 2D spatial extents, currently lacking the ability to deal with 3D tracking in the confines of monocular video. Unfortunately, advances in 3D tracking mainly rely on expensive sensor inputs, e.g., point clouds, depth measurements, radar. Absence of language counterpart for the outputs of these mildly democratized sensors in the literature also hinders VLT expansion to 3D tracking. Addressing that, we make the first attempt towards extending VLT to 3D tracking based on monocular video. We present a comprehensive framework, introducing (i) the Monocular-Video-based 3D Visual Language Tracking (Mono3DVLT) task, (ii) a large-scale dataset for the task, called Mono3DVLT-V2X, and (iii) a customized neural model for the task. Our dataset is carefully curated, leveraging a Large Langauge Model (LLM) followed by human verification, composing natural language descriptions for 79,158 video sequences aiming at single object tracking, providing 2D and 3D bounding box annotations. Our neural model, termed Mono3DVLT-MT, is the first targeted approach for the Mono3DVLT task. Comprising the pipeline of multi-modal feature extractor, visual-language encoder, tracking decoder and a tracking head, our model sets a strong baseline for the task on Mono3DVLT-V2X. Experimental results show that our method significantly outperforms existing techniques on the Mono3DVLT-V2X dataset. Our dataset and code are available in https://github.com/hongkai-wei/Mono3DVLT.

We investigate the generation of minority samples using pretrained text-to-image (T2I) latent diffusion models. Minority instances, in the context of T2I generation, can be defined as ones living on low-density regions of text-conditional data distributions. They are valuable for various applications of modern T2I generators, such as data augmentation and creative AI. Unfortunately, existing pretrained T2I diffusion models primarily focus on high-density regions, largely due to the influence of guided samplers (like CFG) that are essential for high-quality generation. To address this, we present a novel framework to counter the high-density-focus of T2I diffusion models. Specifically, we first develop an online prompt optimization framework that encourages emergence of desired properties during inference while preserving semantic contents of user-provided prompts. We subsequently tailor this generic prompt optimizer into a specialized solver that promotes generation of minority features by incorporating a carefully-crafted likelihood objective. Extensive experiments conducted across various types of T2I models demonstrate that our approach significantly enhances the capability to produce high-quality minority instances compared to existing samplers. Code is available at https://github.com/soobin-um/MinorityPrompt.

Real-world videos consist of sequences of events. Generating such sequences with precise temporal control is infeasible with existing video generators that rely on a single paragraph of text as input. When tasked with generating multiple events described using a single prompt, such methods often ignore some of the events or fail to arrange them in the correct order. To address this limitation, we present MinT, a multi-event video generator with temporal control. Our key insight is to bind each event to a specific period in the generated video, which allows the model to focus on one event at a time. To enable time-aware interactions between event captions and video tokens, we design a time-based positional encoding method, dubbed ReRoPE. This encoding helps to guide the cross-attention operation. By fine-tuning a pre-trained video diffusion transformer on temporally grounded data, our approach produces coherent videos with smoothly connected events. For the first time in the literature, our model offers control over the timing of events in generated videos. Extensive experiments demonstrate that MinT outperforms existing commercial and open-source models by a large margin.

Text serves as the key control signal in video generation due to its narrative nature. To render text descriptions into video clips, current video diffusion models borrow features from text encoders yet struggle with limited text comprehension. The recent success of large language models (LLMs) showcases the power of decoder-only transformers, which offers three clear benefits for text-to-video (T2V) generation, namely, precise text understanding resulting from the superior scalability, imagination beyond the input text enabled by next token prediction, and flexibility to prioritize user interests through instruction tuning. Nevertheless, the feature distribution gap emerging from the two different text modeling paradigms hinders the direct use of LLMs in established T2V models. This work addresses this challenge with Mimir, an end-to-end training framework featuring a carefully tailored token fuser to harmonize the outputs from text encoders and LLMs. Such a design allows the T2V model to fully leverage learned video priors while capitalizing on the text-related capability of LLMs. Extensive quantitative and qualitative results demonstrate the effectiveness of our approach in generating high-quality videos with excellent text comprehension, especially when processing short captions and managing shifting motions. The code and models will be made publicly available.