PyTorchKR Papers

최신 논문

총 1,880편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2410.10774

Cavia: Camera-controllable Multi-view Video Diffusion with View-Integrated Attention

In recent years there have been remarkable breakthroughs in image-to-video generation. However, the 3D consistency and camera controllability of generated frames have remained unsolved. Recent studies have attempted to incorporate camera control into the generation process, but their results are often limited to simple trajectories or lack the ability to generate consistent videos from multiple distinct camera paths for the same scene. To address these limitations, we introduce Cavia, a novel framework for camera-controllable, multi-view video generation, capable of converting an input image into multiple spatiotemporally consistent videos. Our framework extends the spatial and temporal attention modules into view-integrated attention modules, improving both viewpoint and temporal consistency. This flexible design allows for joint training with diverse curated data sources, including scene-level static videos, object-level synthetic multi-view dynamic videos, and real-world monocular dynamic videos. To our best knowledge, Cavia is the first of its kind that allows the user to precisely specify camera motion while obtaining object motion. Extensive experiments demonstrate that Cavia surpasses state-of-the-art methods in terms of geometric consistency and perceptual quality. Project Page: https://ir1d.github.io/Cavia/

arXiv:2410.10751

DragEntity: Trajectory Guided Video Generation using Entity and Positional Relationships

In recent years, diffusion models have achieved tremendous success in the field of video generation, with controllable video generation receiving significant attention. However, existing control methods still face two limitations: Firstly, control conditions (such as depth maps, 3D Mesh) are difficult for ordinary users to obtain directly. Secondly, it's challenging to drive multiple objects through complex motions with multiple trajectories simultaneously. In this paper, we introduce DragEntity, a video generation model that utilizes entity representation for controlling the motion of multiple objects. Compared to previous methods, DragEntity offers two main advantages: 1) Our method is more user-friendly for interaction because it allows users to drag entities within the image rather than individual pixels. 2) We use entity representation to represent any object in the image, and multiple objects can maintain relative spatial relationships. Therefore, we allow multiple trajectories to control multiple objects in the image with different levels of complexity simultaneously. Our experiments validate the effectiveness of DragEntity, demonstrating its excellent performance in fine-grained control in video generation.

arXiv:2410.10816공식 코드 · Papers with Code

LVD-2M: A Long-take Video Dataset with Temporally Dense Captions

The efficacy of video generation models heavily depends on the quality of their training datasets. Most previous video generation models are trained on short video clips, while recently there has been increasing interest in training long video generation models directly on longer videos. However, the lack of such high-quality long videos impedes the advancement of long video generation. To promote research in long video generation, we desire a new dataset with four key features essential for training long video generation models: (1) long videos covering at least 10 seconds, (2) long-take videos without cuts, (3) large motion and diverse contents, and (4) temporally dense captions. To achieve this, we introduce a new pipeline for selecting high-quality long-take videos and generating temporally dense captions. Specifically, we define a set of metrics to quantitatively assess video quality including scene cuts, dynamic degrees, and semantic-level quality, enabling us to filter high-quality long-take videos from a large amount of source videos. Subsequently, we develop a hierarchical video captioning pipeline to annotate long videos with temporally-dense captions. With this pipeline, we curate the first long-take video dataset, LVD-2M, comprising 2 million long-take videos, each covering more than 10 seconds and annotated with temporally dense captions. We further validate the effectiveness of LVD-2M by fine-tuning video generation models to generate long videos with dynamic motions. We believe our work will significantly contribute to future research in long video generation.

arXiv:2410.10802

Boosting Camera Motion Control for Video Diffusion Transformers

Recent advancements in diffusion models have significantly enhanced the quality of video generation. However, fine-grained control over camera pose remains a challenge. While U-Net-based models have shown promising results for camera control, transformer-based diffusion models (DiT)-the preferred architecture for large-scale video generation - suffer from severe degradation in camera motion accuracy. In this paper, we investigate the underlying causes of this issue and propose solutions tailored to DiT architectures. Our study reveals that camera control performance depends heavily on the choice of conditioning methods rather than camera pose representations that is commonly believed. To address the persistent motion degradation in DiT, we introduce Camera Motion Guidance (CMG), based on classifier-free guidance, which boosts camera control by over 400%. Additionally, we present a sparse camera control pipeline, significantly simplifying the process of specifying camera poses for long videos. Our method universally applies to both U-Net and DiT models, offering improved camera control for video generation tasks.

arXiv:2410.10076공식 코드 · Papers with Code

VideoAgent: Self-Improving Video Generation

Video generation has been used to generate visual plans for controlling robotic systems. Given an image observation and a language instruction, previous work has generated video plans which are then converted to robot controls to be executed. However, a major bottleneck in leveraging video generation for control lies in the quality of the generated videos, which often suffer from hallucinatory content and unrealistic physics, resulting in low task success when control actions are extracted from the generated videos. While scaling up dataset and model size provides a partial solution, integrating external feedback is both natural and essential for grounding video generation in the real world. With this observation, we propose VideoAgent for self-improving generated video plans based on external feedback. Instead of directly executing the generated video plan, VideoAgent first refines the generated video plans using a novel procedure which we call self-conditioning consistency, utilizing feedback from a pretrained vision-language model (VLM). As the refined video plan is being executed, VideoAgent collects additional data from the environment to further improve video plan generation. Experiments in simulated robotic manipulation from MetaWorld and iTHOR show that VideoAgent drastically reduces hallucination, thereby boosting success rate of downstream manipulation tasks. We further illustrate that VideoAgent can effectively refine real-robot videos, providing an early indicator that robotics can be an effective tool in grounding video generation in the physical world.

arXiv:2410.08534

Quality Prediction of AI Generated Images and Videos: Emerging Trends and Opportunities

The advent of AI has influenced many aspects of human life, from self-driving cars and intelligent chatbots to text-based image and video generation models capable of creating realistic images and videos based on user prompts (text-to-image, image-to-image, and image-to-video). AI-based methods for image and video super resolution, video frame interpolation, denoising, and compression have already gathered significant attention and interest in the industry and some solutions are already being implemented in real-world products and services. However, to achieve widespread integration and acceptance, AI-generated and enhanced content must be visually accurate, adhere to intended use, and maintain high visual quality to avoid degrading the end user's quality of experience (QoE). One way to monitor and control the visual "quality" of AI-generated and -enhanced content is by deploying Image Quality Assessment (IQA) and Video Quality Assessment (VQA) models. However, most existing IQA and VQA models measure visual fidelity in terms of "reconstruction" quality against a pristine reference content and were not designed to assess the quality of "generative" artifacts. To address this, newer metrics and models have recently been proposed, but their performance evaluation and overall efficacy have been limited by datasets that were too small or otherwise lack representative content and/or distortion capacity; and by performance measures that can accurately report the success of an IQA/VQA model for "GenAI". This paper examines the current shortcomings and possibilities presented by AI-generated and enhanced image and video content, with a particular focus on end-user perceived quality. Finally, we discuss open questions and make recommendations for future work on the "GenAI" quality assessment problems, towards further progressing on this interesting and relevant field of research.

arXiv:2410.08260

Koala-36M: A Large-scale Video Dataset Improving Consistency between Fine-grained Conditions and Video Content

As visual generation technologies continue to advance, the scale of video datasets has expanded rapidly, and the quality of these datasets is critical to the performance of video generation models. We argue that temporal splitting, detailed captions, and video quality filtering are three key factors that determine dataset quality. However, existing datasets exhibit various limitations in these areas. To address these challenges, we introduce Koala-36M, a large-scale, high-quality video dataset featuring accurate temporal splitting, detailed captions, and superior video quality. The core of our approach lies in improving the consistency between fine-grained conditions and video content. Specifically, we employ a linear classifier on probability distributions to enhance the accuracy of transition detection, ensuring better temporal consistency. We then provide structured captions for the splitted videos, with an average length of 200 words, to improve text-video alignment. Additionally, we develop a Video Training Suitability Score (VTSS) that integrates multiple sub-metrics, allowing us to filter high-quality videos from the original corpus. Finally, we incorporate several metrics into the training process of the generation model, further refining the fine-grained conditions. Our experiments demonstrate the effectiveness of our data processing pipeline and the quality of the proposed Koala-36M dataset. Our dataset and code will be released at https://koala36m.github.io/.

arXiv:2410.07194

Technical Report: Competition Solution For Modelscope-Sora

This report presents the approach adopted in the Modelscope-Sora challenge, which focuses on fine-tuning data for video generation models. The challenge evaluates participants' ability to analyze, clean, and generate high-quality datasets for video-based text-to-video tasks under specific computational constraints. The provided methodology involves data processing techniques such as video description generation, filtering, and acceleration. This report outlines the procedures and tools utilized to enhance the quality of training data, ensuring improved performance in text-to-video generation models.

arXiv:2410.07659공식 코드 · Papers with Code

MotionAura: Generating High-Quality and Motion Consistent Videos using Discrete Diffusion

The spatio-temporal complexity of video data presents significant challenges in tasks such as compression, generation, and inpainting. We present four key contributions to address the challenges of spatiotemporal video processing. First, we introduce the 3D Mobile Inverted Vector-Quantization Variational Autoencoder (3D-MBQ-VAE), which combines Variational Autoencoders (VAEs) with masked token modeling to enhance spatiotemporal video compression. The model achieves superior temporal consistency and state-of-the-art (SOTA) reconstruction quality by employing a novel training strategy with full frame masking. Second, we present MotionAura, a text-to-video generation framework that utilizes vector-quantized diffusion models to discretize the latent space and capture complex motion dynamics, producing temporally coherent videos aligned with text prompts. Third, we propose a spectral transformer-based denoising network that processes video data in the frequency domain using the Fourier Transform. This method effectively captures global context and long-range dependencies for high-quality video generation and denoising. Lastly, we introduce a downstream task of Sketch Guided Video Inpainting. This task leverages Low-Rank Adaptation (LoRA) for parameter-efficient fine-tuning. Our models achieve SOTA performance on a range of benchmarks. Our work offers robust frameworks for spatiotemporal modeling and user-driven video content manipulation. We will release the code, datasets, and models in open-source.

arXiv:2410.08184

Scaling Laws For Diffusion Transformers

Diffusion transformers (DiT) have already achieved appealing synthesis and scaling properties in content recreation, e.g., image and video generation. However, scaling laws of DiT are less explored, which usually offer precise predictions regarding optimal model size and data requirements given a specific compute budget. Therefore, experiments across a broad range of compute budgets, from 1e17 to 6e18 FLOPs are conducted to confirm the existence of scaling laws in DiT for the first time. Concretely, the loss of pretraining DiT also follows a power-law relationship with the involved compute. Based on the scaling law, we can not only determine the optimal model size and required data but also accurately predict the text-to-image generation loss given a model with 1B parameters and a compute budget of 1e21 FLOPs. Additionally, we also demonstrate that the trend of pre-training loss matches the generation performances (e.g., FID), even across various datasets, which complements the mapping from compute to synthesis quality and thus provides a predictable benchmark that assesses model performance and data quality at a reduced cost.

arXiv:2410.08151공식 코드 · Papers with Code

Progressive Autoregressive Video Diffusion Models

Current frontier video diffusion models have demonstrated remarkable results at generating high-quality videos. However, they can only generate short video clips, normally around 10 seconds or 240 frames, due to computation limitations during training. Existing methods naively achieve autoregressive long video generation by directly placing the ending of the previous clip at the front of the attention window as conditioning, which leads to abrupt scene changes, unnatural motion, and error accumulation. In this work, we introduce a more natural formulation of autoregressive long video generation by revisiting the noise level assumption in video diffusion models. Our key idea is to 1. assign the frames with per-frame, progressively increasing noise levels rather than a single noise level and 2. denoise and shift the frames in small intervals rather than all at once. This allows for smoother attention correspondence among frames with adjacent noise levels, larger overlaps between the attention windows, and better propagation of information from the earlier to the later frames. Video diffusion models equipped with our progressive noise schedule can autoregressively generate long videos with much improved fidelity compared to the baselines and minimal quality degradation over time. We present the first results on text-conditioned 60-second (1440 frames) long video generation at a quality close to frontier models. Code and video results are available at https://desaixie.github.io/pa-vdm/.

arXiv:2410.07718공식 코드 · Papers with Code

Hallo2: Long-Duration and High-Resolution Audio-Driven Portrait Image Animation

Recent advances in latent diffusion-based generative models for portrait image animation, such as Hallo, have achieved impressive results in short-duration video synthesis. In this paper, we present updates to Hallo, introducing several design enhancements to extend its capabilities. First, we extend the method to produce long-duration videos. To address substantial challenges such as appearance drift and temporal artifacts, we investigate augmentation strategies within the image space of conditional motion frames. Specifically, we introduce a patch-drop technique augmented with Gaussian noise to enhance visual consistency and temporal coherence over long duration. Second, we achieve 4K resolution portrait video generation. To accomplish this, we implement vector quantization of latent codes and apply temporal alignment techniques to maintain coherence across the temporal dimension. By integrating a high-quality decoder, we realize visual synthesis at 4K resolution. Third, we incorporate adjustable semantic textual labels for portrait expressions as conditional inputs. This extends beyond traditional audio cues to improve controllability and increase the diversity of the generated content. To the best of our knowledge, Hallo2, proposed in this paper, is the first method to achieve 4K resolution and generate hour-long, audio-driven portrait image animations enhanced with textual prompts. We have conducted extensive experiments to evaluate our method on publicly available datasets, including HDTF, CelebV, and our introduced "Wild" dataset. The experimental results demonstrate that our approach achieves state-of-the-art performance in long-duration portrait video animation, successfully generating rich and controllable content at 4K resolution for duration extending up to tens of minutes. Project page https://fudan-generative-vision.github.io/hallo2

arXiv:2410.07763

HARIVO: Harnessing Text-to-Image Models for Video Generation

We present a method to create diffusion-based video models from pretrained Text-to-Image (T2I) models. Recently, AnimateDiff proposed freezing the T2I model while only training temporal layers. We advance this method by proposing a unique architecture, incorporating a mapping network and frame-wise tokens, tailored for video generation while maintaining the diversity and creativity of the original T2I model. Key innovations include novel loss functions for temporal smoothness and a mitigating gradient sampling technique, ensuring realistic and temporally consistent video generation despite limited public video data. We have successfully integrated video-specific inductive biases into the architecture and loss functions. Our method, built on the frozen StableDiffusion model, simplifies training processes and allows for seamless integration with off-the-shelf models like ControlNet and DreamBooth. project page: https://kwonminki.github.io/HARIVO

arXiv:2410.07151

FaceVid-1K: A Large-Scale High-Quality Multiracial Human Face Video Dataset

Generating talking face videos from various conditions has recently become a highly popular research area within generative tasks. However, building a high-quality face video generation model requires a well-performing pre-trained backbone, a key obstacle that universal models fail to adequately address. Most existing works rely on universal video or image generation models and optimize control mechanisms, but they neglect the evident upper bound in video quality due to the limited capabilities of the backbones, which is a result of the lack of high-quality human face video datasets. In this work, we investigate the unsatisfactory results from related studies, gather and trim existing public talking face video datasets, and additionally collect and annotate a large-scale dataset, resulting in a comprehensive, high-quality multiracial face collection named \textbf{FaceVid-1K}. Using this dataset, we craft several effective pre-trained backbone models for face video generation. Specifically, we conduct experiments with several well-established video generation models, including text-to-video, image-to-video, and unconditional video generation, under various settings. We obtain the corresponding performance benchmarks and compared them with those trained on public datasets to demonstrate the superiority of our dataset. These experiments also allow us to investigate empirical strategies for crafting domain-specific video generation tasks with cost-effective settings. We will make our curated dataset, along with the pre-trained talking face video generation models, publicly available as a resource contribution to hopefully advance the research field.

arXiv:2410.05322

Noise Crystallization and Liquid Noise: Zero-shot Video Generation using Image Diffusion Models

Although powerful for image generation, consistent and controllable video is a longstanding problem for diffusion models. Video models require extensive training and computational resources, leading to high costs and large environmental impacts. Moreover, video models currently offer limited control of the output motion. This paper introduces a novel approach to video generation by augmenting image diffusion models to create sequential animation frames while maintaining fine detail. These techniques can be applied to existing image models without training any video parameters (zero-shot) by altering the input noise in a latent diffusion model. Two complementary methods are presented. Noise crystallization ensures consistency but is limited to large movements due to reduced latent embedding sizes. Liquid noise trades consistency for greater flexibility without resolution limitations. The core concepts also allow other applications such as relighting, seamless upscaling, and improved video style transfer. Furthermore, an exploration of the VAE embedding used for latent diffusion models is performed, resulting in interesting theoretical insights such as a method for human-interpretable latent spaces.

arXiv:2410.05591공식 코드 · Papers with Code

TweedieMix: Improving Multi-Concept Fusion for Diffusion-based Image/Video Generation

Despite significant advancements in customizing text-to-image and video generation models, generating images and videos that effectively integrate multiple personalized concepts remains a challenging task. To address this, we present TweedieMix, a novel method for composing customized diffusion models during the inference phase. By analyzing the properties of reverse diffusion sampling, our approach divides the sampling process into two stages. During the initial steps, we apply a multiple object-aware sampling technique to ensure the inclusion of the desired target objects. In the later steps, we blend the appearances of the custom concepts in the de-noised image space using Tweedie's formula. Our results demonstrate that TweedieMix can generate multiple personalized concepts with higher fidelity than existing methods. Moreover, our framework can be effortlessly extended to image-to-video diffusion models, enabling the generation of videos that feature multiple personalized concepts. Results and source code are in our anonymous project page.

arXiv:2410.05363공식 코드 · Papers with Code

Towards World Simulator: Crafting Physical Commonsense-Based Benchmark for Video Generation

Text-to-video (T2V) models like Sora have made significant strides in visualizing complex prompts, which is increasingly viewed as a promising path towards constructing the universal world simulator. Cognitive psychologists believe that the foundation for achieving this goal is the ability to understand intuitive physics. However, the capacity of these models to accurately represent intuitive physics remains largely unexplored. To bridge this gap, we introduce PhyGenBench, a comprehensive \textbf{Phy}sics \textbf{Gen}eration \textbf{Ben}chmark designed to evaluate physical commonsense correctness in T2V generation. PhyGenBench comprises 160 carefully crafted prompts across 27 distinct physical laws, spanning four fundamental domains, which could comprehensively assesses models' understanding of physical commonsense. Alongside PhyGenBench, we propose a novel evaluation framework called PhyGenEval. This framework employs a hierarchical evaluation structure utilizing appropriate advanced vision-language models and large language models to assess physical commonsense. Through PhyGenBench and PhyGenEval, we can conduct large-scale automated assessments of T2V models' understanding of physical commonsense, which align closely with human feedback. Our evaluation results and in-depth analysis demonstrate that current models struggle to generate videos that comply with physical commonsense. Moreover, simply scaling up models or employing prompt engineering techniques is insufficient to fully address the challenges presented by PhyGenBench (e.g., dynamic scenarios). We hope this study will inspire the community to prioritize the learning of physical commonsense in these models beyond entertainment applications. We will release the data and codes at https://github.com/OpenGVLab/PhyGenBench

arXiv:2410.05954공식 코드 · Papers with Code

Pyramidal Flow Matching for Efficient Video Generative Modeling

Video generation requires modeling a vast spatiotemporal space, which demands significant computational resources and data usage. To reduce the complexity, the prevailing approaches employ a cascaded architecture to avoid direct training with full resolution. Despite reducing computational demands, the separate optimization of each sub-stage hinders knowledge sharing and sacrifices flexibility. This work introduces a unified pyramidal flow matching algorithm. It reinterprets the original denoising trajectory as a series of pyramid stages, where only the final stage operates at the full resolution, thereby enabling more efficient video generative modeling. Through our sophisticated design, the flows of different pyramid stages can be interlinked to maintain continuity. Moreover, we craft autoregressive video generation with a temporal pyramid to compress the full-resolution history. The entire framework can be optimized in an end-to-end manner and with a single unified Diffusion Transformer (DiT). Extensive experiments demonstrate that our method supports generating high-quality 5-second (up to 10-second) videos at 768p resolution and 24 FPS within 20.7k A100 GPU training hours. All code and models will be open-sourced at https://pyramid-flow.github.io.

arXiv:2410.05799공식 코드 · Papers with Code

SeeClear: Semantic Distillation Enhances Pixel Condensation for Video Super-Resolution

Diffusion-based Video Super-Resolution (VSR) is renowned for generating perceptually realistic videos, yet it grapples with maintaining detail consistency across frames due to stochastic fluctuations. The traditional approach of pixel-level alignment is ineffective for diffusion-processed frames because of iterative disruptions. To overcome this, we introduce SeeClear--a novel VSR framework leveraging conditional video generation, orchestrated by instance-centric and channel-wise semantic controls. This framework integrates a Semantic Distiller and a Pixel Condenser, which synergize to extract and upscale semantic details from low-resolution frames. The Instance-Centric Alignment Module (InCAM) utilizes video-clip-wise tokens to dynamically relate pixels within and across frames, enhancing coherency. Additionally, the Channel-wise Texture Aggregation Memory (CaTeGory) infuses extrinsic knowledge, capitalizing on long-standing semantic textures. Our method also innovates the blurring diffusion process with the ResShift mechanism, finely balancing between sharpness and diffusion effects. Comprehensive experiments confirm our framework's advantage over state-of-the-art diffusion-based VSR techniques. The code is available: https://github.com/Tang1705/SeeClear-NeurIPS24.

arXiv:2410.05677

T2V-Turbo-v2: Enhancing Video Generation Model Post-Training through Data, Reward, and Conditional Guidance Design

In this paper, we focus on enhancing a diffusion-based text-to-video (T2V) model during the post-training phase by distilling a highly capable consistency model from a pretrained T2V model. Our proposed method, T2V-Turbo-v2, introduces a significant advancement by integrating various supervision signals, including high-quality training data, reward model feedback, and conditional guidance, into the consistency distillation process. Through comprehensive ablation studies, we highlight the crucial importance of tailoring datasets to specific learning objectives and the effectiveness of learning from diverse reward models for enhancing both the visual quality and text-video alignment. Additionally, we highlight the vast design space of conditional guidance strategies, which centers on designing an effective energy function to augment the teacher ODE solver. We demonstrate the potential of this approach by extracting motion guidance from the training datasets and incorporating it into the ODE solver, showcasing its effectiveness in improving the motion quality of the generated videos with the improved motion-related metrics from VBench and T2V-CompBench. Empirically, our T2V-Turbo-v2 establishes a new state-of-the-art result on VBench, with a Total score of 85.13, surpassing proprietary systems such as Gen-3 and Kling.