Found object drawing is a creative art form incorporating everyday objects into imaginative images, offering a refreshing and unique way to express ideas. However, for many people, creating this type of work can be challenging due to difficulties in generating creative ideas and finding suitable reference images to help translate their ideas onto paper. Based on the findings of a formative study, we propose GenFODrawing, a creativity support tool to help users create diverse found object drawings. Our system provides AI-driven textual and visual inspirations, and enhances controllability through sketch-based and box-conditioned image generation, enabling users to create personalized outputs. We conducted a user study with twelve participants to compare GenFODrawing, to a baseline condition where the participants completed the creative tasks using their own desired approaches without access to our system. The study demonstrated that GenFODrawing, enabled easier exploration of diverse ideas, greater agency and control through the creative process, and higher creativity support compared to the baseline. A further open-ended study demonstrated the system's usability and expressiveness, and all participants found the creative process engaging.

Hand-drawn character animation is a vibrant research area in computer graphics and presents unique challenges in achieving geometric consistency while conveying expressive motion details. Traditional skeletal animation methods maintain geometric consistency but often struggle with complex non-rigid elements like flowing hair and skirts, resulting in unnatural deformation and missing secondary dynamics. In contrast, video diffusion models effectively synthesize physically plausible dynamics, but exhibit real-human-like characteristics and geometric distortions when applied to stylized drawings due to the domain gap. In this work, we propose a novel hybrid animation system that integrates the strengths of skeletal animation and video diffusion priors. The core idea is to first generate coarse images from characters retargeted with skeletal animations for geometric consistency guidance, and then enhance these images in terms of texture details and secondary dynamics using video diffusion priors. We formulate the enhancement of coarse images as an inpainting task and propose a domain-adapted diffusion model to refine user-masked regions requiring improvement, particularly those involving secondary dynamics. To further enhance motion realism, we propose a Secondary Dynamics Injection (SDI) strategy during the denoising process to incorporate latent features from a pre-trained diffusion model enriched with human motion priors. Additionally, to address unnatural deformation artifacts caused by the integrated hair-body geometry in low-poly single-mesh character modeling, we introduce a Hair Layering Modeling (HLM) technique that employs segmentation maps to separate hair from the body in implicit fields, enabling more natural animation of challenging long-hair characters. Through extensive experiments, we demonstrate that our system outperforms state-of-the-art works in both quantitative and qualitative evaluations. Please refer to our project page (https://lordliang.github.io/From-Rigging-to-Waving) for the code and data for our method.

Recent advancements in human fashion video generation have transformed the field, producing various promising effects. Existing methods mainly focus on pose control but lack the ability to achieve sketch-based control, largely due to the absence of appearance-consistent and shape-varying knowledge in existing datasets. Moreover, the necessity of sequential structure inputs to control video generation hinders real-world applications. To address these limitations, we introduce Sketch2HumanVideo, an approach that, for the first time, achieves sketch-controllable human video generation with three conditions: temporally sparse sketches, a spatially sparse pose sequence, and a reference appearance image. Our key contribution is a sparse sketch encoder, which takes the first two conditions as input, enabling precise and multi-view control of shape motion. To provide the above knowledge, we leverage the expertise of two pretrained models to synthesize a dataset comprising shape-varying yet appearance-consistent examples for model training. Furthermore, we introduce an enlarging-and-resampling scheme to enhance high-frequency details of local regions in resource-constrained scenarios, thereby promoting the generation of realistic videos. Through qualitative and quantitative experiments, our method showcases superior performance to state-of-the-art approaches and flexible control.

The experimental evaluations and a perceptual user study show that our proposed method outperforms the existing 2D and 3D animation methods and generates high-quality 3D animations from a single character drawing.Please refer to our project page (https://lordliang.github.io/DrawingSpinUp)for the code and generated animations.

Defocus deblurring is a challenging task due to the spatially varying nature of defocus blur. While deep learning approach shows great promise in solving image restoration problems, defocus deblurring demands accurate training data that consists of all-in-focus and defocus image pairs, which is difficult to collect. Naive two-shot capturing cannot achieve pixel-wise correspondence between the defocused and all-in-focus image pairs. Synthetic aperture of light fields is suggested to be a more reliable way to generate accurate image pairs. However, the defocus blur generated from light field data is different from that of the images captured with a traditional digital camera. In this paper, we propose a novel deep defocus deblurring network that leverages the strength and overcomes the shortcoming of light fields. We first train the network on a light field-generated dataset for its highly accurate image correspondence. Then, we fine-tune the network using feature loss on another dataset collected by the two-shot method to alleviate the differences between the defocus blur exists in the two domains. This strategy is proved to be highly effective and able to achieve the state-of-the-art performance both quantitatively and qualitatively on multiple test sets. Extensive ablation studies have been conducted to analyze the effect of each network module to the final performance.

We present a deep neural network called the light field generative adversarial network (LFGAN) that synthesizes a 4D light field from a single 2D RGB image. We generate light fields using a single image super-resolution (SISR) technique based on two important observations. First, the small baseline gives rise to the high similarity between the full light field image and each sub-aperture view. Second, the occlusion edge at any spatial coordinate of a sub-aperture view has the same orientation as the occlusion edge at the corresponding angular patch, implying that the occlusion information in the angular domain can be inferred from the sub-aperture local information. We employ the Wasserstein GAN with gradient penalty (WGAN-GP) to learn the color and geometry information from the light field datasets. The network can generate a plausible 4D light field comprising 8×8 angular views from a single sub-aperture 2D image. We propose new loss terms, namely epipolar plane image (EPI) and brightness regularization (BRI) losses, as well as a novel multi-stage training framework to feed the loss terms at different time to generate superior light fields. The EPI loss can reinforce the network to learn the geometric features of the light fields, and the BRI loss can preserve the brightness consistency across different sub-aperture views. Two datasets have been used to evaluate our method: in addition to an existing light field dataset capturing scenes of flowers and plants, we have built a large dataset of toy animals consisting of 2,100 light fields captured with a plenoptic camera. We have performed comprehensive ablation studies to evaluate the effects of individual loss terms and the multi-stage training strategy, and have compared LFGAN to other state-of-the-art techniques. Qualitative and quantitative evaluation demonstrates that LFGAN can effectively estimate complex occlusions and geometry in challenging scenes, and outperform other existing techniques.

This paper presents a novel volumetric display based on projection on a non-planer and reconfigurable fog screen. Unlike conventional fog projection systems which produce 2D images on flat screens, our display scatters different parts of the projected image at different depth levels, thus allowing volumetric data to be displayed in the real 3D space. We constructed the fog screen with a 2D array of nozzles that are individually switchable, while the switching pattern is tightly synchronized with the video content. Our system is superior to many existing approaches at many levels. First, our display does not require head tracking, glasses or head-mounted devices while allowing high resolution, full color 3D image to be observed from wide viewing angles by many people at the same time. As compare with various existing approaches, our system is relatively easy to setup and low cost. Most importantly, our immaterial, mid-air display allows users to directly touch and manipulate virtual objects in 3D under marker-free and barrier-free settings which opens up immense tangible and creative interaction possibilities. In this paper, we provide the details of display mechanism and design prototype, as well as a constrained optimization problem to find the projection distance that can maximize the display resolution. A number of real display examples will demonstrate the performance of the proposed system.

Autonomous deployment of mobile agents for coverage enhancement is an important issue in wireless sensor networks. The major challenge lies in the requirement of efficient distributed and localized computing. In addition, managing the coverage of heterogeneous sensing model is complicated due to the diversity of sensing ranges and the irregularity of coverage holes. This paper presents two distributed algorithms for maximizing the sensing coverage in heterogeneous sensor networks. The first algorithm is based on a circle packing technique. We prove the uniqueness of a circle packing up to a given triangulation and boundary conditions, thus the designated coverage layout can be achieved by controlling the boundary conditions. In the second algorithm, we give a formulation of virtual forces among sensor nodes to reduce redundant overlaps and avoid coverage holes. We prove that these virtual forces always give a quasioptimal local coverage. This method is applicable for deployment of sensor nodes in, not only an open field, but also any bounded field of interest and/or in the presence of obstacles. Numerical simulations are showed and these examples verify that the proposed algorithms always yield sensor deployments of wide coverage and collision free motions among sensor nodes. The proposed strategies utilize only the local information about a sensor node and its neighbors, thus providing distributed, efficient and scalable solutions to the deployment problem.

This paper addresses the problem of deploying a set of mobile sensor nodes of heterogeneous sensing ranges to give a large and connected coverage. A novel deployment algorithm based on the circle packing technique is given. It iteratively enhances the coverage area of the sensor network from an initial random deployment, while guarantees the absence of coverage hole and obstacle avoidance. The sensing area of each node is modeled as a circular disc while its radius is bounded by the corresponding sensing range limit. The problem of placing these circular discs to cover a field is intuitively transformed to the circle packing problem: given the specified combinatorics of tangency patterns of n circles, find the label R denoting the radii of these circles. Since a unique packing exists for any given set of triangulations and boundary conditions, we can always find the minimum sensing range required for every interior node to satisfy such packing conditions. Though an extension from tangency packing to overlap packing, the interstices among triples (which represent coverage holes) can be eliminated. We have proven that the maximum global scaling factor to vanish all possible interstices is alpha = 3^((1/2)/2. Based on a number of numerical simulations, we have verified that the proposed algorithm always yields sensor deployments of wide coverage and minimize the sensing ranges required for every interior sensing node to satisfy the packing and boundary conditions.

This paper presents a novel algorithm, called ISOGRID (isometric grid-based algorithm), for autonomous deployment of mobile sensor networks. Upon an initial random placement of sensor nodes, the algorithm iteratively computes node movements to enhance sensing coverage and avoid obstacles while ensuring sensor connectivity. The principle is to redeploy the sensor nodes such that the communication graph approximates the layout of an isometric grid. Based on a number of simulation experiments, we have verified that the proposed algorithm always yields sensor deployments of wide coverage and desired topologies while ensuring collision-free motions of robots and promising wireless communication among sensor nodes. As the algorithm runs in a decentralized framework, it is computationally efficient and scalable. We also suggest another deployment algorithm, MEC (minimum enclosing circle-based algorithm), as an improvement to Minimax presented in (G. L. Wang, et al., March 2004) and extensively utilize it in the simulation examples to make comparison with ISOGRID

A complete and efficient algorithm is proposed for searching form-closure grasps of n hard fingers on the surface of a three-dimensional object represented by discrete points. Both frictional and frictionless cases are considered. This algorithm starts to search a form-closure grasp from a randomly selected grasp using an efficient local search procedure until encountering a local minimum. The local search procedure employs the powerful ray-shooting technique to search in the direction of reducing the distance between the convex hull corresponding to the grasp and the origin of the wrench space. When the distance reaches a local minimum in the local search procedure, the algorithm decomposes the problem into a few subproblems in subsets of the points according to the existence conditions of form-closure grasps. A search tree whose root represents the original problem is employed to perform the searching process. The subproblems are represented as children of the root node and the same procedure is recursively applied to the children. It is proved that the search tree generates O(KlnK/n) nodes in case a from-closure grasp exists, where K is the number of the local minimum points of the distance in the grasp space and n is the number of fingers. Compared to the exhaustive search, this algorithm is more efficient, and, compared to other heuristic algorithms, the proposed algorithm is complete in the discrete domain. The efficiency of this algorithm is demonstrated by numerical examples.

A complete and efficient algorithm is proposed for searching form-closure grasps of n-hard fingers on 3-D objects represented by discrete points. Both frictional and frictionless cases are considered. This algorithm starts to search a form-closure grasp front a random grasp using an efficient local search procedure until encountering a local minimum. The local procedure is based on the powerful ray-shooting technique and searches in the direction of reducing the distance between the convex hull corresponding to the grasp and the origin of the wrench space. When the distance reaches a local minimum value, the algorithm decomposes the problem into sub-problems according to the existence conditions of form-closure grasps. A search tree whose root represents the original problem is employed to guide the searching process. The sub-problems are represented as children of the root node and the same procedure is recursively applied to the children. Theoretical analysis has been conducted on completeness and computational complexity of the algorithm. The efficiency of this algorithm is demonstrated by numerical examples.

This paper presents an algorithm to calculate the fingertip positions, which can enhance the closure property for a n-finger grasp and ensure the kinematic feasibility. The algorithm starts by selecting a set of initial fingertip positions within the workspace of the robotic hand and on the surface of the object. If the currently selected grasp does not form closure, then the origin of the wrench space lies out side of the convex hull of the primitive contact wrenches. In this case the adjacent positions of each grip point can be generated to form a candidate set of grasps and the most promising one will be adopted based on the measure of a grasp from being a form closure. The measure is defined by considering the distance between the convex hull and the origin. The algorithm is executed iteratively until the origin is eventually contained by the convex hull. Finally, we illustrate the efficiency of the proposed algorithm with an implementation of three numerical examples.

Distributed virtual environments allow users at different geographical locations to share and interact within a common virtual environment via a local network or through the Internet. To deliver a good performance for such applications, we need to address several issues in different research disciplines. First, we must be able to model virtual objects effectively. The recently developed multi-resolution techniques for object modeling are of great value here, since they are capable of simplifying the object models and therefore reducing the time to render them. This may greatly reduce the demand for rendering performance on the client machines. Second, with the constraint of the limited bandwidth of the Internet, we need to reduce the response time by reducing the amount of data requested over the network. Caching of suitable object models of high affinity will reduce the amount of data requested over the network for a faster response time. Prefetching object models by predicting those ...