arXiv subjects

Qiuyu Wang

Publications and source records attributed to Qiuyu Wang.

36 records · Page 2Linked to original sources

arXiv2025

Time Is a Feature: Exploiting Temporal Dynamics in Diffusion Language Models

Diffusion large language models (dLLMs) generate text through iterative denoising, yet current decoding strategies discard rich intermediate predictions in favor of the final output. Our work here reveals a critical phenomenon, temporal oscillation, where correct answers often emerge in the middle process, but are overwritten in later denoising steps. To address this issue, we introduce two complementary methods that exploit temporal consistency: 1) Temporal Self-Consistency Voting, a training-free, test-time decoding strategy that aggregates predictions across denoising steps to select the most consistent output; and 2) a post-training method termed Temporal Consistency Reinforcement, which uses Temporal Semantic Entropy (TSE), a measure of semantic stability across intermediate predictions, as a reward signal to encourage stable generations. Empirical results across multiple benchmarks demonstrate the effectiveness of our approach. Using the negative TSE reward alone, we observe a remarkable average improvement of 24.7% on the Countdown dataset over an existing dLLM. Combined with the accuracy reward, we achieve absolute gains of 2.0% on GSM8K, 4.3% on MATH500, 6.6% on SVAMP, and 25.3% on Countdown, respectively. Our findings underscore the untapped potential of temporal dynamics in dLLMs and offer two simple yet effective tools to harness them.

cs.CL↗

arXiv2025

Calligrapher: Freestyle Text Image Customization

We introduce Calligrapher, a novel diffusion-based framework that innovatively integrates advanced text customization with artistic typography for digital calligraphy and design applications. Addressing the challenges of precise style control and data dependency in typographic customization, our framework incorporates three key technical contributions. First, we develop a self-distillation mechanism that leverages the pre-trained text-to-image generative model itself alongside the large language model to automatically construct a style-centric typography benchmark. Second, we introduce a localized style injection framework via a trainable style encoder, which comprises both Qformer and linear layers, to extract robust style features from reference images. An in-context generation mechanism is also employed to directly embed reference images into the denoising process, further enhancing the refined alignment of target styles. Extensive quantitative and qualitative evaluations across diverse fonts and design contexts confirm Calligrapher's accurate reproduction of intricate stylistic details and precise glyph positioning. By automating high-quality, visually consistent typography, Calligrapher surpasses traditional models, empowering creative practitioners in digital art, branding, and contextual typographic design.

cs.CV↗

arXiv2025

LeviTor: 3D Trajectory Oriented Image-to-Video Synthesis

The intuitive nature of drag-based interaction has led to its growing adoption for controlling object trajectories in image-to-video synthesis. Still, existing methods that perform dragging in the 2D space usually face ambiguity when handling out-of-plane movements. In this work, we augment the interaction with a new dimension, i.e., the depth dimension, such that users are allowed to assign a relative depth for each point on the trajectory. That way, our new interaction paradigm not only inherits the convenience from 2D dragging, but facilitates trajectory control in the 3D space, broadening the scope of creativity. We propose a pioneering method for 3D trajectory control in image-to-video synthesis by abstracting object masks into a few cluster points. These points, accompanied by the depth information and the instance information, are finally fed into a video diffusion model as the control signal. Extensive experiments validate the effectiveness of our approach, dubbed LeviTor, in precisely manipulating the object movements when producing photo-realistic videos from static images. Our code is available at: https://github.com/ant-research/LeviTor.

cs.CV↗

arXiv2025

MagicQuill: An Intelligent Interactive Image Editing System

Image editing involves a variety of complex tasks and requires efficient and precise manipulation techniques. In this paper, we present MagicQuill, an integrated image editing system that enables swift actualization of creative ideas. Our system features a streamlined yet functionally robust interface, allowing for the articulation of editing operations (e.g., inserting elements, erasing objects, altering color) with minimal input. These interactions are monitored by a multimodal large language model (MLLM) to anticipate editing intentions in real time, bypassing the need for explicit prompt entry. Finally, we apply a powerful diffusion prior, enhanced by a carefully learned two-branch plug-in module, to process editing requests with precise control. Experimental results demonstrate the effectiveness of MagicQuill in achieving high-quality image edits. Please visit https://magic-quill.github.io to try out our system.

cs.CV↗

arXiv2025

Learning Naturally Aggregated Appearance for Efficient 3D Editing

Neural radiance fields, which represent a 3D scene as a color field and a density field, have demonstrated great progress in novel view synthesis yet are unfavorable for editing due to the implicitness. This work studies the task of efficient 3D editing, where we focus on editing speed and user interactivity. To this end, we propose to learn the color field as an explicit 2D appearance aggregation, also called canonical image, with which users can easily customize their 3D editing via 2D image processing. We complement the canonical image with a projection field that maps 3D points onto 2D pixels for texture query. This field is initialized with a pseudo canonical camera model and optimized with offset regularity to ensure the naturalness of the canonical image. Extensive experiments on different datasets suggest that our representation, dubbed AGAP, well supports various ways of 3D editing (e.g., stylization, instance segmentation, and interactive drawing). Our approach demonstrates remarkable efficiency by being at least 20 times faster per edit compared to existing NeRF-based editing methods. Project page is available at https://felixcheng97.github.io/AGAP/.

cs.CV↗

arXiv2025

AniDoc: Animation Creation Made Easier

The production of 2D animation follows an industry-standard workflow, encompassing four essential stages: character design, keyframe animation, in-betweening, and coloring. Our research focuses on reducing the labor costs in the above process by harnessing the potential of increasingly powerful generative AI. Using video diffusion models as the foundation, AniDoc emerges as a video line art colorization tool, which automatically converts sketch sequences into colored animations following the reference character specification. Our model exploits correspondence matching as an explicit guidance, yielding strong robustness to the variations (e.g., posture) between the reference character and each line art frame. In addition, our model could even automate the in-betweening process, such that users can easily create a temporally consistent animation by simply providing a character image as well as the start and end sketches. Our code is available at: https://yihao-meng.github.io/AniDoc_demo.

cs.CV↗

arXiv2025

Edicho: Consistent Image Editing in the Wild

As a verified need, consistent editing across in-the-wild images remains a technical challenge arising from various unmanageable factors, like object poses, lighting conditions, and photography environments. Edicho steps in with a training-free solution based on diffusion models, featuring a fundamental design principle of using explicit image correspondence to direct editing. Specifically, the key components include an attention manipulation module and a carefully refined classifier-free guidance (CFG) denoising strategy, both of which take into account the pre-estimated correspondence. Such an inference-time algorithm enjoys a plug-and-play nature and is compatible to most diffusion-based editing methods, such as ControlNet and BrushNet. Extensive results demonstrate the efficacy of Edicho in consistent cross-image editing under diverse settings. We will release the code to facilitate future studies.

cs.CV↗

arXiv2024

CoDeF: Content Deformation Fields for Temporally Consistent Video Processing

We present the content deformation field CoDeF as a new type of video representation, which consists of a canonical content field aggregating the static contents in the entire video and a temporal deformation field recording the transformations from the canonical image (i.e., rendered from the canonical content field) to each individual frame along the time axis. Given a target video, these two fields are jointly optimized to reconstruct it through a carefully tailored rendering pipeline. We advisedly introduce some regularizations into the optimization process, urging the canonical content field to inherit semantics (e.g., the object shape) from the video. With such a design, CoDeF naturally supports lifting image algorithms for video processing, in the sense that one can apply an image algorithm to the canonical image and effortlessly propagate the outcomes to the entire video with the aid of the temporal deformation field. We experimentally show that CoDeF is able to lift image-to-image translation to video-to-video translation and lift keypoint detection to keypoint tracking without any training. More importantly, thanks to our lifting strategy that deploys the algorithms on only one image, we achieve superior cross-frame consistency in processed videos compared to existing video-to-video translation approaches, and even manage to track non-rigid objects like water and smog. Project page can be found at https://qiuyu96.github.io/CoDeF/.

cs.CV↗

arXiv2024

Framer: Interactive Frame Interpolation

We propose Framer for interactive frame interpolation, which targets producing smoothly transitioning frames between two images as per user creativity. Concretely, besides taking the start and end frames as inputs, our approach supports customizing the transition process by tailoring the trajectory of some selected keypoints. Such a design enjoys two clear benefits. First, incorporating human interaction mitigates the issue arising from numerous possibilities of transforming one image to another, and in turn enables finer control of local motions. Second, as the most basic form of interaction, keypoints help establish the correspondence across frames, enhancing the model to handle challenging cases (e.g., objects on the start and end frames are of different shapes and styles). It is noteworthy that our system also offers an "autopilot" mode, where we introduce a module to estimate the keypoints and refine the trajectory automatically, to simplify the usage in practice. Extensive experimental results demonstrate the appealing performance of Framer on various applications, such as image morphing, time-lapse video generation, cartoon interpolation, etc. The code, the model, and the interface will be released to facilitate further research.

cs.CV↗

arXiv2024

Real-time 3D-aware Portrait Editing from a Single Image

This work presents 3DPE, a practical method that can efficiently edit a face image following given prompts, like reference images or text descriptions, in a 3D-aware manner. To this end, a lightweight module is distilled from a 3D portrait generator and a text-to-image model, which provide prior knowledge of face geometry and superior editing capability, respectively. Such a design brings two compelling advantages over existing approaches. First, our method achieves real-time editing with a feedforward network (i.e., ~0.04s per image), over 100x faster than the second competitor. Second, thanks to the powerful priors, our module could focus on the learning of editing-related variations, such that it manages to handle various types of editing simultaneously in the training phase and further supports fast adaptation to user-specified customized types of editing during inference (e.g., with ~5min fine-tuning per style).

cs.CV↗

arXiv2024

InFusion: Inpainting 3D Gaussians via Learning Depth Completion from Diffusion Prior

3D Gaussians have recently emerged as an efficient representation for novel view synthesis. This work studies its editability with a particular focus on the inpainting task, which aims to supplement an incomplete set of 3D Gaussians with additional points for visually harmonious rendering. Compared to 2D inpainting, the crux of inpainting 3D Gaussians is to figure out the rendering-relevant properties of the introduced points, whose optimization largely benefits from their initial 3D positions. To this end, we propose to guide the point initialization with an image-conditioned depth completion model, which learns to directly restore the depth map based on the observed image. Such a design allows our model to fill in depth values at an aligned scale with the original depth, and also to harness strong generalizability from largescale diffusion prior. Thanks to the more accurate depth completion, our approach, dubbed InFusion, surpasses existing alternatives with sufficiently better fidelity and efficiency under various complex scenarios. We further demonstrate the effectiveness of InFusion with several practical applications, such as inpainting with user-specific texture or with novel object insertion.

cs.CV↗

arXiv2024

Uncertainty-driven and Adversarial Calibration Learning for Epicardial Adipose Tissue Segmentation

Epicardial adipose tissue (EAT) is a type of visceral fat that can secrete large amounts of adipokines to affect the myocardium and coronary arteries. EAT volume and density can be used as independent risk markers measurement of volume by noninvasive magnetic resonance images is the best method of assessing EAT. However, segmenting EAT is challenging due to the low contrast between EAT and pericardial effusion and the presence of motion artifacts. we propose a novel feature latent space multilevel supervision network (SPDNet) with uncertainty-driven and adversarial calibration learning to enhance segmentation for more accurate EAT volume estimation. The network first addresses the blurring of EAT edges due to the medical images in the open medical environments with low quality or out-of-distribution by modeling the uncertainty as a Gaussian distribution in the feature latent space, which using its Bayesian estimation as a regularization constraint to optimize SwinUNETR. Second, an adversarial training strategy is introduced to calibrate the segmentation feature map and consider the multi-scale feature differences between the uncertainty-guided predictive segmentation and the ground truth segmentation, synthesizing the multi-scale adversarial loss directly improves the ability to discriminate the similarity between organizations. Experiments on both the cardiac public MRI dataset (ACDC) and the real-world clinical cohort EAT dataset show that the proposed network outperforms mainstream models, validating that uncertainty-driven and adversarial calibration learning can be used to provide additional information for modeling multi-scale ambiguities.

eess.IV↗

arXiv2023

GenDeF: Learning Generative Deformation Field for Video Generation

We offer a new perspective on approaching the task of video generation. Instead of directly synthesizing a sequence of frames, we propose to render a video by warping one static image with a generative deformation field (GenDeF). Such a pipeline enjoys three appealing advantages. First, we can sufficiently reuse a well-trained image generator to synthesize the static image (also called canonical image), alleviating the difficulty in producing a video and thereby resulting in better visual quality. Second, we can easily convert a deformation field to optical flows, making it possible to apply explicit structural regularizations for motion modeling, leading to temporally consistent results. Third, the disentanglement between content and motion allows users to process a synthesized video through processing its corresponding static image without any tuning, facilitating many applications like video editing, keypoint tracking, and video segmentation. Both qualitative and quantitative results on three common video generation benchmarks demonstrate the superiority of our GenDeF method.

cs.CV↗

arXiv2023

An inverse analysis of fluid flow through granular media using differentiable lattice Boltzmann method

Inverse modeling of fluid flow through porous soils and reservoir rocks enables accurate determination of permeability and seepage properties critical for applications such as contaminant filtration, stability assessments, and optimization of hydrocarbon recovery. However, the solution to the inverse problem is ill-posed and sensitive to noise in measurements. Direct simulation of flow through granular media only provides forward predictions rather than inverse characterization. We present an effective method for solving inverse analysis of fluid flow through granular media. The key objectives are accurately determining boundary conditions and characterizing the physical properties of the granular media (e.g., permeability) and fluid viscosity based on the flow state. We develop a fully differentiable lattice Boltzmann Method (LBM) using Automatic Differentiation (AD) to optimize the gradients of the input features with respect to the observations. We implement a GPU-capable AD-LBM using the Domain-Specific Taichi programming language by employing lightweight tape gradients for end-to-end backpropagation. For complex flow paths in porous media, our AD-LBM approach accurately estimates the boundary conditions leading to observed steady-state velocity fields and consequently derives macro-scale permeability and fluid viscosity. Our method demonstrates significant advantages in prediction accuracy and computational efficiency, offering a powerful tool for solving inverse fluid flow problems in various applications.

physics.flu-dyn↗

arXiv2023

Benchmarking and Analyzing 3D-aware Image Synthesis with a Modularized Codebase

Despite the rapid advance of 3D-aware image synthesis, existing studies usually adopt a mixture of techniques and tricks, leaving it unclear how each part contributes to the final performance in terms of generality. Following the most popular and effective paradigm in this field, which incorporates a neural radiance field (NeRF) into the generator of a generative adversarial network (GAN), we build a well-structured codebase, dubbed Carver, through modularizing the generation process. Such a design allows researchers to develop and replace each module independently, and hence offers an opportunity to fairly compare various approaches and recognize their contributions from the module perspective. The reproduction of a range of cutting-edge algorithms demonstrates the availability of our modularized codebase. We also perform a variety of in-depth analyses, such as the comparison across different types of point feature, the necessity of the tailing upsampler in the generator, the reliance on the camera pose prior, etc., which deepen our understanding of existing methods and point out some further directions of the research work. We release code and models at https://github.com/qiuyu96/Carver to facilitate the development and evaluation of this field.

cs.CV↗

arXiv2023

Multiphase lattice Boltzmann modeling of cyclic water retention behavior in unsaturated sand based on X-ray Computed Tomography

The water retention curve (WRC) defines the relationship between matric suction and saturation and is a key function for determining the hydro-mechanical behavior of unsaturated soils. We investigate possible microscopic origins of the water retention behavior of granular soils using both Computed Tomography (CT) experiment and multiphase lattice Boltzmann Method (LBM). We conduct a CT experiment on Hamburg sand to obtain its WRC and then run LBM simulations based on the CT grain skeleton. The multiphase LBM simulations capture the hysteresis and pore-scale behaviors of WRC observed in the CT experiment. Using LBM, we observe that the spatial distribution and morphology of gas clusters varies between drainage and imbibition paths and is the underlying source of the hysteresis. During drainage, gas clusters congregate at the grain surface; the local suction increases when gas clusters enter through small pore openings and decreases when gas clusters enter through large pore openings. Whereas, during imbibition, gas clusters disperse in the liquid; the local suction decreases uniformly. Large pores empty first during drainage and small pores fill first during imbibition. The pore-based WRC shows that an increase in pore size causes a decrease in suction during drainage and imbibition, and an increase in hysteresis.

physics.geo-ph↗

arXiv2021

Effect Of Slope Angle On The Runout Evolution of Granular Column Collapse for Varying Initial Volumes

In nature, submarine slope failures usually carry thousands of cubic-meters of sediments across extremely long distances and cause tsunamis and damages to offshore structures. This paper uses the granular column collapse experiment to investigate the effect of slope angle on the runout behavior of submarine granular landslides for different initial volumes. A two-dimensional coupled lattice Boltzman and discrete element method (LBM-DEM) approach is adopted for numerically modeling the granular column collapse. Columns with four different slope angles and six different volumes are modelled under both dry and submerged conditions. The effects of hydrodynamic interactions, including the generation of excess pore pressures, hydroplaning, and drag forces and formation of turbulent vortices, are used to explain the difference in the runout behavior of the submerged columns compared to the dry columns. The results show that at any given slope angle, there is a threshold volume above which the submerged columns have a larger final runout compared to their dry counterpart, and this threshold volume decreases with slope angle.

physics.geo-ph↗

arXiv2020

Effect of Initial Volume on the Run-Out Behavior of Submerged Granular Columns

Submarine landslides transport thousands of cubic meters of sediment across continental shelves even at slopes as low as 1° and can cause significant casualty and damage to infrastructure. The run-out mechanism in a submarine landslide is affected by factors such as the initial packing density, permeability, slope angle, and initial volume. While past studies have focused on the influence of density, permeability, and slope angle on the granular column collapse, the impact of volume on the run-out characteristics has not been investigated. This study aims to understand how the initial volume affects the run-out using a two-dimensional coupled lattice Boltzman and discrete element (LBM-DEM) method. The coupled LBM-DEM approach allows simulating fluid flow at the pore-scale resolution to understand the grain-scale mechanisms driving the complex continuum-scale response in the granular column collapse. For submerged granular column collapse, the run-out mechanism is heavily influenced by the interaction between the grains and the surrounding fluid. The development of negative pore pressures during shearing and hydrodynamic drag forces inhibit the flow. On the other hand, entrainment of water resulting in hydroplaning enhances the flow. With an increase in volume, the interaction between the grains and the surrounding fluid varies, causing changes in the run-out behavior. For smaller volumes, the forces inhibiting the underwater flow predominates, resulting in shorter run-outs than their dry counterparts. At large volumes, hydroplaning results in larger run-out than the dry cases, despite the inhibiting effects of drag forces and negative pore pressures.

physics.geo-ph↗