Generating multiple distinct subjects remains a challenge for existing text-to-image diffusion models. Complex prompts often lead to subject leakage, causing inaccuracies in quantities, attributes, and visual features. Preventing leakage among subjects necessitates knowledge of each subject’s spatial location. Recent methods provide these spatial locations via an external layout control. However, enforcing such a prescribed layout often conflicts with the innate layout dictated by the sampled initial noise, leading to misalignment with the model’s prior. In this work, we introduce a new approach that predicts a spatial layout aligned with the prompt, derived from the initial noise, and refines it throughout the denoising process. By relying on this noise-induced layout, we avoid conflicts with externally imposed layouts and better preserve the model’s prior. Our method employs a small neural network to predict and refine the evolving noise-induced layout at each denoising step, ensuring clear boundaries between subjects while maintaining consistency. Experimental results show that this noise-aligned strategy achieves improved text-image alignment and more stable multi-subject generation compared to existing layout-guided techniques, while preserving the rich diversity of the model’s original distribution.
论文检索
输入标题、作者或关键词，从 9,256 篇学术成果中精准定位
Speech-driven 3D facial animation plays a key role in applications such as virtual avatars, gaming, and digital content creation. While existing methods have made significant progress in achieving accurate lip synchronization and generating basic emotional expressions, they often struggle to capture and effectively transfer nuanced performance styles. We propose a novel example-based generation framework that conditions a latent diffusion model on a reference style clip to produce highly expressive and temporally coherent facial animations. To address the challenge of accurately adhering to the style reference, we introduce a novel conditioning mechanism called style basis, which extracts key poses from the reference and additively guides the diffusion generation process to fit the style without compromising lip synchronization quality. This approach enables the model to capture subtle stylistic cues while ensuring that the generated animations align closely with the input speech. Extensive qualitative, quantitative, and perceptual evaluations demonstrate the effectiveness of our method in faithfully reproducing the desired style while achieving superior lip synchronization across various speech scenarios.
Recent advancements in large vision-language models have enabled highly expressive and diverse vector sketch generation. However, state-of-the-art methods rely on a time-consuming optimization process involving repeated feedback from a pretrained model to determine stroke placement. Consequently, despite producing impressive sketches, these methods are limited in practical applications. In this work, we introduce SwiftSketch, a diffusion model for image-conditioned vector sketch generation that can produce high-quality sketches in less than a second. SwiftSketch operates by progressively denoising stroke control points sampled from a Gaussian distribution. Its transformer-decoder architecture is designed to effectively handle the discrete nature of vector representation and capture the inherent global dependencies between strokes. To train SwiftSketch, we construct a synthetic dataset of image-sketch pairs, addressing the limitations of existing sketch datasets, which are often created by non-artists and lack professional quality. For generating these synthetic sketches, we introduce ControlSketch, a method that enhances SDS-based techniques by incorporating precise spatial control through a depth-aware ControlNet. We demonstrate that SwiftSketch generalizes across diverse concepts, efficiently producing sketches that combine high fidelity with a natural and visually appealing style.
Recently, 3D generation methods have shown their powerful ability to automate 3D model creation. However, most 3D generation methods only rely on an input image or a text prompt to generate a 3D model, which lacks the control of each component of the generated 3D model. Any modifications of the input image lead to an entire regeneration of the 3D models. In this paper, we introduce a new method called CMD that generates a 3D model from an input image while enabling flexible local editing of each component of the 3D model. In CMD, we formulate the 3D generation as a conditional multiview diffusion model, which takes the existing or known parts as conditions and generates the edited or added components. This conditional multiview diffusion model not only allows the generation of 3D models part by part but also enables local editing of 3D models according to the local revision of the input image without changing other 3D parts. Extensive experiments are conducted to demonstrate that CMD decomposes a complex 3D generation task into multiple components, improving the generation quality. Meanwhile, CMD enables efficient and flexible local editing of a 3D model by just editing one rendered image.
Recent advancements in generative models have significantly propelled 3D scene editing. While existing methods excel at text-guided texture modifications for 3D representations like 3D Gaussian Splatting (3DGS), they struggle with geometric transformations (e.g., rotating a character’s head) and lack precise spatial control over edits due to the inherent ambiguity of language-driven guidance. To address these limitations, we introduce DYG, a 3D drag-based editing framework for 3DGS. Users intuitively define editing regions using 3D masks and specify desired transformations through pairs of control points. DYG integrates the implicit triplane representation to establish the geometric scaffold of editing results, effectively overcoming suboptimal editing outcomes caused by the sparsity of 3DGS in the desired editing regions. Additionally, we incorporate a drag-based Latent Diffusion Model through the proposed Drag-SDS loss, enabling flexible, multi-view consistent, and fine-grained editing. Extensive experiments demonstrate that DYG enables effective drag-based editing, outperforming other baselines in terms of editing effect and quality. Additional results are available on our project page: https://quyans.github.io/Drag-Your-Gaussian/.
Recent advancements in 3D generation techniques have simplified the tedious manual process of 3D asset production. Among these methods, 3D native latent diffusion models are particularly effective in generating high-quality geometric details. However, achieving local enhancement and editing of the generated 3D models remains a challenge due to the limited understanding of the relationship between text,images,and 3D in terms of local semantics and feature space.We explore and reveal the characteristics of the native 3D latent space, make it decomposable and low-rank, thereby enabling efficient and effective learning for multimodal local alignment. Based on this, we introduce RELATE3D, a novel approach that combines a Refocusing Adapter with part-to-latent correspondence guided training for precise local enhancement and part-level editing of 3D geometry. The Refocusing Adapter incorporates partial image and caption signals, and, combined with part-to-latent mapping, directs modifications to the relevant latent dimensions during latent diffusion process. We validate the effectiveness of our approach through extensive experiments and ablation studies, showcasing the capabilities of our generative local enhancement and editing process, as well as global refinement.
Point-based representations have consistently played a vital role in geometric data structures. Most point cloud learning and processing methods typically leverage the unordered and unconstrained nature to represent the underlying geometry of 3D shapes. However, how to extract meaningful structural information from unstructured point cloud distributions and transform them into semantically meaningful point distributions remains an under-explored problem. We present PDT, a novel framework for point distribution transformation with diffusion models. Given a set of input points, PDT learns to transform the point set from its original geometric distribution into a target distribution that is semantically meaningful. Our method utilizes diffusion models with novel architecture and learning strategy, which effectively correlates the source and the target distribution through a denoising process. Through extensive experiments, we show that our method successfully transforms input point clouds into various forms of structured outputs - ranging from surface-aligned keypoints, and inner sparse joints to continuous feature lines. The results showcase our framework’s ability to capture both geometric and semantic features, offering a powerful tool for various 3D geometry processing tasks where structured point distributions are desired. Code will be available at this link: link.
The comic production industry requires reference-based line art colorization with high accuracy, efficiency, contextual consistency, and flexible control.A comic page often involves diverse characters, objects, and backgrounds, which complicates the coloring process. Despite advancements in diffusion models for image generation, their application in line art colorization remains limited, facing challenges related to handling extensive reference images, time-consuming inference, and flexible control. We investigate the necessity of extensive contextual image guidance on the quality of line art colorization. To address these challenges, we introduce Cobra, an efficient and versatile method that supports color hints and utilizes over 200 reference images while maintaining low latency. Central to Cobra is a Causal Sparse DiT architecture, which leverages specially designed positional encodings, causal sparse attention, and Key-Value Cache to effectively manage long-context references and ensure color identity consistency. Results demonstrate that Cobra achieves accurate line art colorization through extensive contextual reference, significantly enhancing inference speed and interactivity, thereby meeting critical industrial demands. We release our codes and models on our project page: https://zhuang2002.github.io/Cobra/.
Scalable Vector Graphics (SVGs) are highly favored by designers due to their resolution independence and well-organized layer structure. Although existing text-to-vector (T2V) generation methods can create SVGs from text prompts, they often overlook an important need in practical applications: style customization, which is vital for producing a collection of vector graphics with consistent visual appearance and coherent aesthetics. Extending existing T2V methods for style customization poses certain challenges. Optimization-based T2V models can utilize the priors of text-to-image (T2I) models for customization, but struggle with maintaining structural regularity. On the other hand, feed-forward T2V models can ensure structural regularity, yet they encounter difficulties in disentangling content and style due to limited SVG training data. To address these challenges, we propose a novel two-stage style customization pipeline for SVG generation, making use of the advantages of both feed-forward T2V models and T2I image priors. In the first stage, we train a T2V diffusion model with a path-level representation to ensure the structural regularity of SVGs while preserving diverse expressive capabilities. In the second stage, we customize the T2V diffusion model to different styles by distilling customized T2I models. By integrating these techniques, our pipeline can generate high-quality and diverse SVGs in custom styles based on text prompts in an efficient feed-forward manner. The effectiveness of our method has been validated through extensive experiments. The project page is https://customsvg.github.io.
We introduce ReStyle3D, a novel framework for scene-level appearance transfer from a single style image to a real-world scene represented by multiple views. The method combines explicit semantic correspondences with multi-view consistency to achieve precise and coherent stylization. Unlike conventional stylization methods that apply a reference style globally, ReStyle3D uses open-vocabulary segmentation to establish dense, instance-level correspondences between the style and real-world images. This ensures that each object is stylized with semantically matched textures. ReStyle3D first transfers the style to a single view using a training-free semantic-attention mechanism in a diffusion model. It then lifts the stylization to additional views via a learned warp-and-refine network guided by monocular depth and pixel-wise correspondences. Experiments show that ReStyle3D consistently outperforms prior methods in structure preservation, perceptual style similarity, and multi-view coherence. User studies further validate its ability to produce photo-realistic, semantically faithful results. Our code, pretrained models, and dataset will be publicly released, to support new applications in interior design, virtual staging, and 3D-consistent stylization. Project page and code at https://restyle3d.github.io/.
Current makeup transfer methods are limited to simple makeup styles, making them difficult to apply in real-world scenarios. In this paper, we introduce Stable-Makeup, a novel diffusion-based makeup transfer method capable of robustly transferring a wide range of real-world makeup, onto user-provided faces. Stable-Makeup is based on a pre-trained diffusion model and utilizes a Detail-Preserving (D-P) makeup encoder to encode makeup details. It also employs content and structural control modules to preserve the content and structural information of the source image. With the aid of our newly added makeup cross-attention layers in U-Net, we can accurately transfer the detailed makeup to the corresponding position in the source image. After content-structure decoupling training, Stable-Makeup can maintain the content and the facial structure of the source image. Moreover, our method has demonstrated strong robustness and generalizability, making it applicable to various tasks such as cross-domain makeup transfer, makeup-guided text-to-image generation, and so on. Extensive experiments have demonstrated that our approach delivers state-of-the-art results among existing makeup transfer methods and exhibits a highly promising with broad potential applications in various related fields.
Video matting has traditionally been limited by the lack of high-quality ground-truth data. Most existing video matting datasets provide only human-annotated imperfect alpha and foreground annotations, which must be composited to background images or videos during the training stage. Thus, the generalization capability of previous methods in real-world scenarios is typically poor. In this work, we propose to solve the problem from two perspectives. First, we emphasize the importance of large-scale pre-training by pursuing diverse synthetic and pseudo-labeled segmentation datasets. We also develop a scalable synthetic data generation pipeline that can render diverse human bodies and fine-grained hairs, yielding around 200 video clips with a 3-second duration for fine-tuning. Second, we introduce a novel video matting approach that can effectively leverage the rich priors from pre-trained video diffusion models. This architecture offers two key advantages. First, strong priors play a critical role in bridging the domain gap between synthetic and real-world scenes. Second, unlike most existing methods that process video matting frame-by-frame and use an independent decoder to aggregate temporal information, our model is inherently designed for video, ensuring strong temporal consistency. We provide a comprehensive quantitative evaluation across three benchmark datasets, demonstrating our approach’s superior performance, and present comprehensive qualitative results in diverse real-world scenes, illustrating the strong generalization capability of our method. Code is available at https://github.com/aim-uofa/GVM
We present LayerFlow, a unified solution for layer-aware video generation. Given per-layer prompts, LayerFlow generates videos for the transparent foreground, clean background, and blended scene. It also supports versatile variants like decomposing a blended video or generating the background for the given foreground and vice versa. Starting from a text-to-video diffusion transformer, we organize the videos of different layers as sub-clips, and leverage layer embeddings to distinguish each clip and the corresponding layer-wise prompts. In this way, we seamlessly support the aforementioned variants in one unified framework. For the lack of high-quality layer-wise training videos, we design a multi-stage training strategy to accommodate static images with high-quality layer annotations. Specifically, we first train the model with low-quality video data. Then, we tune a motion LoRA to make the model compatible with static frames. Afterward, we train the content LoRA on the mixture of image data with high-quality layered images along with copy-pasted video data. During inference, we remove the motion LoRA thus generating smooth videos with desired layers.
We present Mobius, a novel method to generate seamlessly looping videos from text descriptions directly without any user annotations, thereby creating new visual materials for the multi-media presentation. Our method repurposes the pre-trained video latent diffusion model for generating looping videos from text prompts without any training. During inference, we first construct a latent cycle by connecting the starting and ending noise of the videos. Given that the temporal consistency can be maintained by the context of the video diffusion model, we perform multi-frame latent denoising by gradually shifting the first-frame latent to the end in each step. As a result, the denoising context varies in each step while maintaining consistency throughout the inference process. Moreover, the latent cycle in our method can be of any length. This extends our latent-shifting approach to generate seamless looping videos beyond the scope of the video diffusion model’s context. Unlike previous cinemagraphs, the proposed method does not require an image as appearance, which will restrict the motions of the generated results. Instead, our method can produce more dynamic motion and better visual quality. We conduct multiple experiments and comparisons to verify the effectiveness of the proposed method, demonstrating its efficacy in different scenarios. All the code will be made available.
Diffusion models have demonstrated impressive performance in generating high-quality videos from text prompts or images. However, precise control over the video generation process—such as camera manipulation or content editing—remains a significant challenge. Existing methods for controlled video generation are typically limited to a single control type, lacking the flexibility to handle diverse control demands. In this paper, we introduce Diffusion as Shader (DaS), a novel approach that supports multiple video control tasks within a unified architecture. Our key insight is that achieving versatile video control necessitates leveraging 3D control signals, as videos are fundamentally 2D renderings of dynamic 3D content. Unlike prior methods limited to 2D control signals, DaS leverages 3D tracking videos as control inputs, making the video diffusion process inherently 3D-aware. This innovation allows DaS to achieve a wide range of video controls by simply manipulating the 3D tracking videos. A further advantage of using 3D tracking videos is their ability to effectively link frames, significantly enhancing the temporal consistency of the generated videos. With just 3 days of fine-tuning on 8 H800 GPUs using less than 10k videos, DaS demonstrates strong control capabilities across diverse tasks, including mesh-to-video generation, camera control, motion transfer, and object manipulation. More results are available in the supplementary materials.
This paper presents a method that allows users to design cinematic video shots in the context of image-to-video generation. Shot design, a critical aspect of filmmaking, involves meticulously planning both camera movements and object motions in a scene. However, enabling intuitive shot design in modern image-to-video generation systems presents two main challenges: first, effectively capturing user intentions on the motion design, where both camera movements and scene-space object motions must be specified jointly; and second, representing motion information that can be effectively utilized by a video diffusion model to synthesize the image animations. To address these challenges, we introduce MotionCanvas, a method that integrates user-driven controls into image-to-video (I2V) generation models, allowing users to control both object and camera motions in a scene-aware manner. By connecting insights from classical computer graphics and contemporary video generation techniques, we demonstrate the ability to achieve 3D-aware motion control in I2V synthesis without requiring costly 3D-related training data. MotionCanvas enables users to intuitively depict scene-space motion intentions, and translates them into spatiotemporal motion-conditioning signals for video diffusion models. We demonstrate the effectiveness of our method on a wide range of real-world image content and shot-design scenarios, highlighting its potential to enhance the creative workflows in digital content creation and adapt to various image and video editing applications. Code and model weights are at https://motion-canvas25.github.io
In this work, we present CineMaster, a novel framework for 3D-aware and controllable text-to-video generation. Our goal is to empower users with comparable controllability as professional film directors: precise placement of objects within the scene, flexible manipulation of both objects and camera in 3D space, and intuitive layout control over the rendered frames. To achieve this, CineMaster operates in two stages. In the first stage, we design an interactive workflow that allows users to intuitively construct 3D-aware control signals by positioning object bounding boxes and defining camera movements within the 3D space. In the second stage, these control signals—comprising rendered depth maps, camera trajectories, and object class labels—serve as the guidance for a text-to-video diffusion model, ensuring to generate the user-intended video content. Furthermore, to overcome the scarcity of in-the-wild datasets with 3D object motion and camera pose annotations, we carefully establish an automated data annotation pipeline that extracts 3D bounding boxes and camera trajectories from large-scale video data. Extensive qualitative and quantitative experiments demonstrate that CineMaster significantly outperforms existing methods and implements prominent 3D-aware text-to-video generation.
Text-to-stylized human motion generation leverages text descriptions for motion generation with fine-grained style control with respect to a reference motion. However, existing approaches typically rely on supervised style learning with labeled datasets, constraining their adaptability and generalization for effective diverse style control. Additionally, they have not fully explored the temporal correlations between motion, textual descriptions, and style, making it challenging to generate semantically consistent motion with precise style alignment. To address these limitations, we introduce a novel method that integrates unsupervised style from arbitrary references into a text-driven diffusion model to generate semantically consistent stylized human motion. The core innovation lies in leveraging text as a mediator to capture the temporal correspondences between motion and style, enabling the seamless integration of temporally dynamic style into motion features. Specifically, we first train a diffusion model on a text-motion dataset to capture the correlation between motion and text semantics. A style adapter then extracts temporally dynamic style features from reference motions and integrates a novel Semantic-Aware Style Injection (SASI) module to infuse these features into the diffusion model. The SASI module computes the semantic correlation between motion and style features based on text, selectively incorporating style features that align with motion content, ensuring semantic consistency and precise style alignment. Our style adapter does not require a labeled style dataset for training, enhancing adaptability and generalization of style control. Extensive evaluations show that our method outperforms previous approaches in terms of semantic consistency and style expressivity. Our webpage, https://fivezerojun.github.io/stylization.github.io/, includes links to the supplementary video and code.
The art of instrument performance stands as a vivid manifestation of human creativity and emotion. Nonetheless, generating instrument performance motions is a highly challenging task, as it requires not only capturing intricate movements but also reconstructing the complex dynamics of the performer-instrument interaction. While existing works primarily focus on modeling partial body motions, we propose Expressive ceLlo performance motion Generation for Audio Rendition (ELGAR), a state-of-the-art diffusion-based framework for whole-body fine-grained instrument performance motion generation solely from audio. To emphasize the interactive nature of the instrument performance, we introduce Hand Interactive Contact Loss (HICL) and Bow Interactive Contact Loss (BICL), which effectively guarantee the authenticity of the interplay. Moreover, to better evaluate whether the generated motions align with the semantic context of the music audio, we design novel metrics specifically for string instrument performance motion generation, including finger-contact distance, bow-string distance, and bowing score. Extensive evaluations and ablation studies are conducted to validate the efficacy of the proposed methods. In addition, we put forward a motion generation dataset SPD-GEN, collated and normalized from the MoCap dataset SPD. As demonstrated, ELGAR has shown great potential in generating instrument performance motions with complicated and fast interactions, which will promote further development in areas such as animation, music education, interactive art creation, etc. Our code and SPD-GEN dataset are available at https://github.com/Qzping/ELGAR.
In this paper, we present a novel neural global illumination approach that enables multi-frequency reflections in dynamic scenes. Our method utilizes object-centric, spatial feature grids as the core framework to model rendering effects implicitly. A lightweight scene query, based on single-bounce ray tracing, is then performed on these feature grids to extract principal and secondary features separately. The principal features can capture a wide range of relatively low-frequency global illumination effects, such as diffuse indirect lighting and reflections on rough surfaces. In contrast, the secondary features can provide sparse scene-specific reflection details, typically with much higher frequencies than the final observed radiance. Inspired by the physical processes of light propagation, we introduce a novel dual-band feature fusion module that seamlessly blends these two types of features, generating fused features capable of modeling multi-frequency rendering effects. Additionally, we propose a two-stage training strategy tailored to accommodate the distinct characteristics of each feature type, significantly enhancing the overall quality and reducing artifacts in the rendered results. Experimental results demonstrate that our method delivers high-quality, multi-frequency dynamic reflections, outperforming state-of-the-art baselines, including path tracing with screen-space neural denoising and other neural global illumination methods.