Showing papers in "ACM Transactions on Graphics in 2021"
TL;DR: This article presents StyleFlow as a simple, effective, and robust solution to both the sub-problems of attribute-conditioned sampling and attribute-controlled editing by formulating conditional exploration as an instance of conditional continuous normalizing flows in the GAN latent space conditioned by attribute features.
Abstract: High-quality, diverse, and photorealistic images can now be generated by unconditional GANs (e.g., StyleGAN). However, limited options exist to control the generation process using (semantic) attributes while stillpreserving the quality of the output. Further, due to the entangled nature of the GAN latent space, performing edits along one attribute can easily result in unwanted changes along other attributes. In this article, in the context of conditional exploration of entangled latent spaces, we investigate the two sub-problems of attribute-conditioned sampling and attribute-controlled editing. We present StyleFlow as a simple, effective, and robust solution to both the sub-problems by formulating conditional exploration as an instance of conditional continuous normalizing flows in the GAN latent space conditioned by attribute features. We evaluate our method using the face and the car latent space of StyleGAN, and demonstrate fine-grained disentangled edits along various attributes on both real photographs and StyleGAN generated images. For example, for faces, we vary camera pose, illumination variation, expression, facial hair, gender, and age. Finally, via extensive qualitative and quantitative comparisons, we demonstrate the superiority of StyleFlow over prior and several concurrent works. Project Page and Video: https://rameenabdal.github.io/StyleFlow.
306 citations
TL;DR: In this article, the authors identify and analyze the existence of a distortioneditability tradeoff and a distortionperception tradeoff within the StyleGAN latent space, and suggest two principles for designing encoders in a manner that allows one to control the proximity of the inversions to regions that StyleGAN was originally trained on.
Abstract: Recently, there has been a surge of diverse methods for performing image editing by employing pre-trained unconditional generators. Applying these methods on real images, however, remains a challenge, as it necessarily requires the inversion of the images into their latent space. To successfully invert a real image, one needs to find a latent code that reconstructs the input image accurately, and more importantly, allows for its meaningful manipulation. In this paper, we carefully study the latent space of StyleGAN, the state-of-the-art unconditional generator. We identify and analyze the existence of a distortion-editability tradeoff and a distortion-perception tradeoff within the StyleGAN latent space. We then suggest two principles for designing encoders in a manner that allows one to control the proximity of the inversions to regions that StyleGAN was originally trained on. We present an encoder based on our two principles that is specifically designed for facilitating editing on real images by balancing these tradeoffs. By evaluating its performance qualitatively and quantitatively on numerous challenging domains, including cars and horses, we show that our inversion method, followed by common editing techniques, achieves superior real-image editing quality, with only a small reconstruction accuracy drop.
158 citations
TL;DR: In this article, the authors proposed a hybrid implicit-explicit network architecture and training strategy that adaptively allocates resources during training and inference based on the local complexity of a signal of interest.
Abstract: Neural representations have emerged as a new paradigm for applications in rendering, imaging, geometric modeling, and simulation. Compared to traditional representations such as meshes, point clouds, or volumes they can be flexibly incorporated into differentiable learning-based pipelines. While recent improvements to neural representations now make it possible to represent signals with fine details at moderate resolutions (e.g., for images and 3D shapes), adequately representing large-scale or complex scenes has proven a challenge. Current neural representations fail to accurately represent images at resolutions greater than a megapixel or 3D scenes with more than a few hundred thousand polygons. Here, we introduce a new hybrid implicit-explicit network architecture and training strategy that adaptively allocates resources during training and inference based on the local complexity of a signal of interest. Our approach uses a multiscale block-coordinate decomposition, similar to a quadtree or octree, that is optimized during training. The network architecture operates in two stages: using the bulk of the network parameters, a coordinate encoder generates a feature grid in a single forward pass. Then, hundreds or thousands of samples within each block can be efficiently evaluated using a lightweight feature decoder. With this hybrid implicit-explicit network architecture, we demonstrate the first experiments that fit gigapixel images to nearly 40 dB peak signal-to-noise ratio. Notably this represents an increase in scale of over 1000X compared to the resolution of previously demonstrated image-fitting experiments. Moreover, our approach is able to represent 3D shapes significantly faster and better than previous techniques; it reduces training times from days to hours or minutes and memory requirements by over an order of magnitude.
134 citations
TL;DR: In this paper, an adversarial RL procedure automatically selects which motion to perform, dynamically interpolating and generalizing from a dataset of unstructured motion clips, without any explicit clip selection or sequencing.
Abstract: Synthesizing graceful and life-like behaviors for physically simulated characters has been a fundamental challenge in computer animation. Data-driven methods that leverage motion tracking are a prominent class of techniques for producing high fidelity motions for a wide range of behaviors. However, the effectiveness of these tracking-based methods often hinges on carefully designed objective functions, and when applied to large and diverse motion datasets, these methods require significant additional machinery to select the appropriate motion for the character to track in a given scenario. In this work, we propose to obviate the need to manually design imitation objectives and mechanisms for motion selection by utilizing a fully automated approach based on adversarial imitation learning. High-level task objectives that the character should perform can be specified by relatively simple reward functions, while the low-level style of the character's behaviors can be specified by a dataset of unstructured motion clips, without any explicit clip selection or sequencing. For example, a character traversing an obstacle course might utilize a task-reward that only considers forward progress, while the dataset contains clips of relevant behaviors such as running, jumping, and rolling. These motion clips are used to train an adversarial motion prior, which specifies style-rewards for training the character through reinforcement learning (RL). The adversarial RL procedure automatically selects which motion to perform, dynamically interpolating and generalizing from the dataset. Our system produces high-quality motions that are comparable to those achieved by state-of-the-art tracking-based techniques, while also being able to easily accommodate large datasets of unstructured motion clips. Composition of disparate skills emerges automatically from the motion prior, without requiring a high-level motion planner or other task-specific annotations of the motion clips. We demonstrate the effectiveness of our framework on a diverse cast of complex simulated characters and a challenging suite of motor control tasks.
122 citations
TL;DR: DECA as discussed by the authors disentangles person-specific details from expression-dependent wrinkles by introducing a detail-consistency loss, which allows to synthesize realistic personspecific wrinkles by controlling expression parameters while keeping personspecific details unchanged.
Abstract: While current monocular 3D face reconstruction methods can recover fine geometric details, they suffer several limitations. Some methods produce faces that cannot be realistically animated because they do not model how wrinkles vary with expression. Other methods are trained on high-quality face scans and do not generalize well to in-the-wild images. We present the first approach that regresses 3D face shape and animatable details that are specific to an individual but change with expression. Our model, DECA (Detailed Expression Capture and Animation), is trained to robustly produce a UV displacement map from a low-dimensional latent representation that consists of person-specific detail parameters and generic expression parameters, while a regressor is trained to predict detail, shape, albedo, expression, pose and illumination parameters from a single image. To enable this, we introduce a novel detail-consistency loss that disentangles person-specific details from expression-dependent wrinkles. This disentanglement allows us to synthesize realistic person-specific wrinkles by controlling expression parameters while keeping person-specific details unchanged. DECA is learned from in-the-wild images with no paired 3D supervision and achieves state-of-the-art shape reconstruction accuracy on two benchmarks. Qualitative results on in-the-wild data demonstrate DECA's robustness and its ability to disentangle identity- and expression-dependent details enabling animation of reconstructed faces. The model and code are publicly available at https://deca.is.tue.mpg.de.
98 citations
TL;DR: In this article, an image-to-image translation method that learns to directly encode real facial images into the latent space of a pre-trained unconditional GAN (e.g., StyleGAN) subject to a given aging shift is presented.
Abstract: The task of age transformation illustrates the change of an individual's appearance over time. Accurately modeling this complex transformation over an input facial image is extremely challenging as it requires making convincing, possibly large changes to facial features and head shape, while still preserving the input identity. In this work, we present an image-to-image translation method that learns to directly encode real facial images into the latent space of a pre-trained unconditional GAN (e.g., StyleGAN) subject to a given aging shift. We employ a pre-trained age regression network to explicitly guide the encoder in generating the latent codes corresponding to the desired age. In this formulation, our method approaches the continuous aging process as a regression task between the input age and desired target age, providing fine-grained control over the generated image. Moreover, unlike approaches that operate solely in the latent space using a prior on the path controlling age, our method learns a more disentangled, non-linear path. Finally, we demonstrate that the end-to-end nature of our approach, coupled with the rich semantic latent space of StyleGAN, allows for further editing of the generated images. Qualitative and quantitative evaluations show the advantages of our method compared to state-of-the-art approaches. Code is available at our project page: https://yuval-alaluf.github.io/SAM.
87 citations
TL;DR: In this paper, the authors propose a new layered neural representation, where each dynamic entity including the environment itself is formulated into a spatio-temporal coherent neural layered radiance representation called ST-NeRF, which achieves the disentanglement of location, deformation and appearance of the dynamic entity in a continuous and self-supervised manner.
Abstract: Generating free-viewpoint videos is critical for immersive VR/AR experience, but recent neural advances still lack the editing ability to manipulate the visual perception for large dynamic scenes. To fill this gap, in this paper, we propose the first approach for editable free-viewpoint video generation for large-scale view-dependent dynamic scenes using only 16 cameras. The core of our approach is a new layered neural representation, where each dynamic entity, including the environment itself, is formulated into a spatio-temporal coherent neural layered radiance representation called ST-NeRF. Such a layered representation supports manipulations of the dynamic scene while still supporting a wide free viewing experience. In our ST-NeRF, we represent the dynamic entity/layer as a continuous function, which achieves the disentanglement of location, deformation as well as the appearance of the dynamic entity in a continuous and self-supervised manner. We propose a scene parsing 4D label map tracking to disentangle the spatial information explicitly and a continuous deform module to disentangle the temporal motion implicitly. An object-aware volume rendering scheme is further introduced for the re-assembling of all the neural layers. We adopt a novel layered loss and motion-aware ray sampling strategy to enable efficient training for a large dynamic scene with multiple performers, Our framework further enables a variety of editing functions, i.e., manipulating the scale and location, duplicating or retiming individual neural layers to create numerous visual effects while preserving high realism. Extensive experiments demonstrate the effectiveness of our approach to achieve high-quality, photo-realistic, and editable free-viewpoint video generation for dynamic scenes.
81 citations
TL;DR: TransPose as discussed by the authors is a DNN-based approach to perform full motion capture (with both global translations and body poses) from only 6 Inertial Measurement Units (IMUs) at over 90 fps.
Abstract: Motion capture is facing some new possibilities brought by the inertial sensing technologies which do not suffer from occlusion or wide-range recordings as vision-based solutions do. However, as the recorded signals are sparse and quite noisy, online performance and global translation estimation turn out to be two key difficulties. In this paper, we present TransPose, a DNN-based approach to perform full motion capture (with both global translations and body poses) from only 6 Inertial Measurement Units (IMUs) at over 90 fps. For body pose estimation, we propose a multi-stage network that estimates leaf-to-full joint positions as intermediate results. This design makes the pose estimation much easier, and thus achieves both better accuracy and lower computation cost. For global translation estimation, we propose a supporting-foot-based method and an RNN-based method to robustly solve for the global translations with a confidence-based fusion technique. Quantitative and qualitative comparisons show that our method outperforms the state-of-the-art learning- and optimization-based methods with a large margin in both accuracy and efficiency. As a purely inertial sensor-based approach, our method is not limited by environmental settings (e.g., fixed cameras), making the capture free from common difficulties such as wide-range motion space and strong occlusion.
64 citations
TL;DR: Wang et al. as discussed by the authors proposed a physics-based cloth simulation in a very high resolution setting, presumably at submillimeter levels with millions of vertices, to meet perceptual precision of our human eyes.
Abstract: In this paper, we study physics-based cloth simulation in a very high resolution setting, presumably at submillimeter levels with millions of vertices, to meet perceptual precision of our human eyes. State-of-the-art simulation techniques, mostly developed for unstructured triangular meshes, can hardly meet this demand due to their large computational costs and memory footprints. We argue that in a very high resolution, it is more plausible to use regular meshes with an underlying grid structure, which can be highly compatible with GPU acceleration like high-resolution images. Based on this idea, we formulate and solve the nonlinear optimization problem for simulating high-resolution wrinkles, by a fast block-based descent method with reduced memory accesses. We also investigate the development of the collision handling component in our system, whose performance benefits greatly from the grid structure. Finally, we explore various issues related to the applications of our system, including initialization for fast convergence and temporal coherence, gathering effects, inflation and stuffing models, and mesh simplification. We can treat our system as a quasistatic wrinkle synthesis tool, run it as a standalone dynamic simulator, or integrate it into a multi-resolution solver as an additional component. The experiment demonstrates the capability, efficiency and flexibility of our system in producing a variety of high-resolution wrinkles effects.
63 citations
TL;DR: In this paper, the authors present a real-time neural radiance caching method for path-traced global illumination, which makes no assumptions about the lighting, geometry, and materials.
Abstract: We present a real-time neural radiance caching method for path-traced global illumination. Our system is designed to handle fully dynamic scenes, and makes no assumptions about the lighting, geometry, and materials. The data-driven nature of our approach sidesteps many difficulties of caching algorithms, such as locating, interpolating, and updating cache points. Since pretraining neural networks to handle novel, dynamic scenes is a formidable generalization challenge, we do away with pretraining and instead achieve generalization via adaptation, i.e. we opt for training the radiance cache while rendering. We employ self-training to provide low-noise training targets and simulate infinite-bounce transport by merely iterating few-bounce training updates. The updates and cache queries incur a mild overhead---about 2.6ms on full HD resolution---thanks to a streaming implementation of the neural network that fully exploits modern hardware. We demonstrate significant noise reduction at the cost of little induced bias, and report state-of-the-art, real-time performance on a number of challenging scenarios.
62 citations
TL;DR: FovVideoVDP as mentioned in this paper is a video difference metric that models the spatial, temporal, and peripheral aspects of perception, which is derived from psychophysical studies of the early visual system, which model spatio-temporal contrast sensitivity, cortical magnification and contrast masking.
Abstract: FovVideoVDP is a video difference metric that models the spatial, temporal, and peripheral aspects of perception. While many other metrics are available, our work provides the first practical treatment of these three central aspects of vision simultaneously. The complex interplay between spatial and temporal sensitivity across retinal locations is especially important for displays that cover a large field-of-view, such as Virtual and Augmented Reality displays, and associated methods, such as foveated rendering. Our metric is derived from psychophysical studies of the early visual system, which model spatio-temporal contrast sensitivity, cortical magnification and contrast masking. It accounts for physical specification of the display (luminance, size, resolution) and viewing distance. To validate the metric, we collected a novel foveated rendering dataset which captures quality degradation due to sampling and reconstruction. To demonstrate our algorithm's generality, we test it on 3 independent foveated video datasets, and on a large image quality dataset, achieving the best performance across all datasets when compared to the state-of-the-art.
TL;DR: In this article, a pose conditioned StyleGAN2 architecture with a clothing segmentation branch is proposed to preserve and synthesize skin color and target body shape while transferring the garment from a different person.
Abstract: Given a pair of images---target person and garment on another person---we automatically generate the target person in the given garment. Previous methods mostly focused on texture transfer via paired data training, while overlooking body shape deformations, skin color, and seamless blending of garment with the person. This work focuses on those three components, while also not requiring paired data training. We designed a pose conditioned StyleGAN2 architecture with a clothing segmentation branch that is trained on images of people wearing garments. Once trained, we propose a new layered latent space interpolation method that allows us to preserve and synthesize skin color and target body shape while transferring the garment from a different person. We demonstrate results on high resolution 512 × 512 images, and extensively compare to state of the art in try-on on both latent space generated and real images.
TL;DR: In this article, a learned, differentiable forward model for compound optics and an alternating proximal optimization method for function compositions with highly varying parameter dimensions for optics, hardware image signal processing (ISP) and neural nets are proposed.
Abstract: Most modern commodity imaging systems we use directly for photography—or indirectly rely on for downstream applications—employ optical systems of multiple lenses that must balance deviations from perfect optics, manufacturing constraints, tolerances, cost, and footprint. Although optical designs often have complex interactions with downstream image processing or analysis tasks, today’s compound optics are designed in isolation from these interactions. Existing optical design tools aim to minimize optical aberrations, such as deviations from Gauss’ linear model of optics, instead of application-specific losses, precluding joint optimization with hardware image signal processing (ISP) and highly parameterized neural network processing. In this article, we propose an optimization method for compound optics that lifts these limitations. We optimize entire lens systems jointly with hardware and software image processing pipelines, downstream neural network processing, and application-specific end-to-end losses. To this end, we propose a learned, differentiable forward model for compound optics and an alternating proximal optimization method that handles function compositions with highly varying parameter dimensions for optics, hardware ISP, and neural nets. Our method integrates seamlessly atop existing optical design tools, such as Zemax. We can thus assess our method across many camera system designs and end-to-end applications. We validate our approach in an automotive camera optics setting—together with hardware ISP post processing and detection—outperforming classical optics designs for automotive object detection and traffic light state detection. For human viewing tasks, we optimize optics and processing pipelines for dynamic outdoor scenarios and dynamic low-light imaging. We outperform existing compartmentalized design or fine-tuning methods qualitatively and quantitatively, across all domain-specific applications tested.
TL;DR: AgileGAN as mentioned in this paper introduces a hierarchical variational autoencoder to ensure the inverse mapped distribution conforms to the original latent Gaussian distribution, while augmenting the original space to a multi-resolution latent space so as to better encode different levels of detail.
Abstract: Portraiture as an art form has evolved from realistic depiction into a plethora of creative styles. While substantial progress has been made in automated stylization, generating high quality stylistic portraits is still a challenge, and even the recent popular Toonify suffers from several artifacts when used on real input images. Such StyleGAN-based methods have focused on finding the best latent inversion mapping for reconstructing input images; however, our key insight is that this does not lead to good generalization to different portrait styles. Hence we propose AgileGAN, a framework that can generate high quality stylistic portraits via inversion-consistent transfer learning. We introduce a novel hierarchical variational autoencoder to ensure the inverse mapped distribution conforms to the original latent Gaussian distribution, while augmenting the original space to a multi-resolution latent space so as to better encode different levels of detail. To better capture attribute-dependent stylization of facial features, we also present an attribute-aware generator and adopt an early stopping strategy to avoid overfitting small training datasets. Our approach provides greater agility in creating high quality and high resolution (1024×1024) portrait stylization models, requiring only a limited number of style exemplars (~100) and short training time (~1 hour). We collected several style datasets for evaluation including 3D cartoons, comics, oil paintings and celebrities. We show that we can achieve superior portrait stylization quality to previous state-of-the-art methods, with comparisons done qualitatively, quantitatively and through a perceptual user study. We also demonstrate two applications of our method, image editing and motion retargeting.
TL;DR: In this paper, the authors proposed a method to assess the similarity between color palettes by sorting colors, which is called dynamic closest color warping (DCCW) and calculates the minimum distance sum between colors and the graph connecting the colors in the other palette.
Abstract: A color palette is one of the simplest and most intuitive descriptors that can be extracted from images or videos. This paper proposes a method to assess the similarity between color palettes by sorting colors. While previous palette similarity measures compare only colors without considering the overall palette combination, we sort palettes to minimize the geometric distance between colors and align them to share a common color tendency. We propose dynamic closest color warping (DCCW) to calculate the minimum distance sum between colors and the graph connecting the colors in the other palette. We evaluate the proposed palette sorting and DCCW with several datasets and demonstrate that DCCW outperforms previous methods in terms of accuracy and computing time. We validate the effectiveness of the proposed sorting technique by conducting a perceptual study, which indicates a clear preference for the results of our approach. We also demonstrate useful applications enabled by DCCW, including palette interpolation, palette navigation, and image recoloring.
TL;DR: In this article, a novel general-purpose Style and WAvelet-based GAN (SWAGAN) is proposed, which incorporates wavelets throughout its generator and discriminator architectures, enforcing a frequency-aware latent representation at every step of the way.
Abstract: In recent years, considerable progress has been made in the visual quality of Generative Adversarial Networks (GANs). Even so, these networks still suffer from degradation in quality for high-frequency content, stemming from a spectrally biased architecture, and similarly unfavorable loss functions. To address this issue, we present a novel general-purpose Style and WAvelet based GAN (SWAGAN) that implements progressive generation in the frequency domain. SWAGAN incorporates wavelets throughout its generator and discriminator architectures, enforcing a frequency-aware latent representation at every step of the way. This approach, designed to directly tackle the spectral bias of neural networks, yields an improvement in the ability to generate medium and high frequency content, including structures which other networks fail to learn. We demonstrate the advantage of our method by integrating it into the SyleGAN2 framework, and verifying that content generation in the wavelet domain leads to more realistic high-frequency content, even when trained for fewer iterations. Furthermore, we verify that our model's latent space retains the qualities that allow StyleGAN to serve as a basis for a multitude of editing tasks, and show that our frequency-aware approach also induces improved high-frequency performance in downstream tasks.
TL;DR: MVP as discussed by the authors is a hybrid representation that combines the completeness of volumetric representations with the efficiency of primitive-based rendering, e.g., point-based or mesh-based methods.
Abstract: Real-time rendering and animation of humans is a core function in games, movies, and telepresence applications. Existing methods have a number of drawbacks we aim to address with our work. Triangle meshes have difficulty modeling thin structures like hair, volumetric representations like Neural Volumes are too low-resolution given a reasonable memory budget, and high-resolution implicit representations like Neural Radiance Fields are too slow for use in real-time applications. We present Mixture of Volumetric Primitives (MVP), a representation for rendering dynamic 3D content that combines the completeness of volumetric representations with the efficiency of primitive-based rendering, e.g., point-based or mesh-based methods. Our approach achieves this by leveraging spatially shared computation with a convolutional architecture and by minimizing computation in empty regions of space with volumetric primitives that can move to cover only occupied regions. Our parameterization supports the integration of correspondence and tracking constraints, while being robust to areas where classical tracking fails, such as around thin or translucent structures and areas with large topological variability. MVP is a hybrid that generalizes both volumetric and primitive-based representations. Through a series of extensive experiments we demonstrate that it inherits the strengths of each, while avoiding many of their limitations. We also compare our approach to several state-of-the-art methods and demonstrate that MVP produces superior results in terms of quality and runtime performance.
TL;DR: In this article, a conditional variational autoencoder is used to generate full-body animation for virtual telepresence with driving signals acquired from minimal sensors placed in the environment and mounted on a VR headset.
Abstract: We present a learning-based method for building driving-signal aware full-body avatars. Our model is a conditional variational autoencoder that can be animated with incomplete driving signals, such as human pose and facial keypoints, and produces a high-quality representation of human geometry and view-dependent appearance. The core intuition behind our method is that better drivability and generalization can be achieved by disentangling the driving signals and remaining generative factors, which are not available during animation. To this end, we explicitly account for information deficiency in the driving signal by introducing a latent space that exclusively captures the remaining information, thus enabling the imputation of the missing factors required during full-body animation, while remaining faithful to the driving signal. We also propose a learnable localized compression for the driving signal which promotes better generalization, and helps minimize the influence of global chance-correlations often found in real datasets. For a given driving signal, the resulting variational model produces a compact space of uncertainty for missing factors that allows for an imputation strategy best suited to a particular application. We demonstrate the efficacy of our approach on the challenging problem of full-body animation for virtual telepresence with driving signals acquired from minimal sensors placed in the environment and mounted on a VR-headset.
TL;DR: In this paper, a neural network is used to regress finger motions from input trajectories of wrists and objects and then predicts a new finger pose for the next frame as an autoregressive model.
Abstract: Natural hand manipulations exhibit complex finger maneuvers adaptive to object shapes and the tasks at hand. Learning dexterous manipulation from data in a brute force way would require a prohibitive amount of examples to effectively cover the combinatorial space of 3D shapes and activities. In this paper, we propose a hand-object spatial representation that can achieve generalization from limited data. Our representation combines the global object shape as voxel occupancies with local geometric details as samples of closest distances. This representation is used by a neural network to regress finger motions from input trajectories of wrists and objects. Specifically, we provide the network with the current finger pose, past and future trajectories, and the spatial representations extracted from these trajectories. The network then predicts a new finger pose for the next frame as an autoregressive model. With a carefully chosen hand-centric coordinate system, we can handle single-handed and two-handed motions in a unified framework. Learning from a small number of primitive shapes and kitchenware objects, the network is able to synthesize a variety of finger gaits for grasping, in-hand manipulation, and bimanual object handling on a rich set of novel shapes and functional tasks. We also demonstrate a live demo of manipulating virtual objects in real-time using a simple physical prop. Our system is useful for offline animation or real-time applications forgiving to a small delay.
TL;DR: ChoreoMaster as discussed by the authors is a production-ready music-driven dance motion synthesis system, which can automatically generate a high-quality dance motion sequence to accompany the input music in terms of style, rhythm and structure.
Abstract: Despite strong demand in the game and film industry, automatically synthesizing high-quality dance motions remains a challenging task. In this paper, we present ChoreoMaster, a production-ready music-driven dance motion synthesis system. Given a piece of music, ChoreoMaster can automatically generate a high-quality dance motion sequence to accompany the input music in terms of style, rhythm and structure. To achieve this goal, we introduce a novel choreography-oriented choreomusical embedding framework, which successfully constructs a unified choreomusical embedding space for both style and rhythm relationships between music and dance phrases. The learned choreomusical embedding is then incorporated into a novel choreography-oriented graph-based motion synthesis framework, which can robustly and efficiently generate high-quality dance motions following various choreographic rules. Moreover, as a production-ready system, ChoreoMaster is sufficiently controllable and comprehensive for users to produce desired results. Experimental results demonstrate that dance motions generated by ChoreoMaster are accepted by professional artists.
TL;DR: In this article, a semi-parametric approach is proposed to learn a neural representation of the light transport of a scene that is embedded in a texture atlas of known but possibly rough geometry.
Abstract: The light transport (LT) of a scene describes how it appears under different lighting conditions from different viewing directions, and complete knowledge of a scene’s LT enables the synthesis of novel views under arbitrary lighting. In this article, we focus on image-based LT acquisition, primarily for human bodies within a light stage setup. We propose a semi-parametric approach for learning a neural representation of the LT that is embedded in a texture atlas of known but possibly rough geometry. We model all non-diffuse and global LT as residuals added to a physically based diffuse base rendering. In particular, we show how to fuse previously seen observations of illuminants and views to synthesize a new image of the same scene under a desired lighting condition from a chosen viewpoint. This strategy allows the network to learn complex material effects (such as subsurface scattering) and global illumination (such as diffuse interreflection), while guaranteeing the physical correctness of the diffuse LT (such as hard shadows). With this learned LT, one can relight the scene photorealistically with a directional light or an HDRI map, synthesize novel views with view-dependent effects, or do both simultaneously, all in a unified framework using a set of sparse observations. Qualitative and quantitative experiments demonstrate that our Neural Light Transport (NLT) outperforms state-of-the-art solutions for relighting and view synthesis, without requiring separate treatments for both problems that prior work requires. The code and data are available at http://nlt.csail.mit.edu.
TL;DR: SketchGNN as mentioned in this paper uses graph convolution and a static-dynamic branching network architecture to extract the features at three levels, i.e., point-level, stroke-level and sketch-level.
Abstract: We introduce SketchGNN, a convolutional graph neural network for semantic segmentation and labeling of freehand vector sketches. We treat an input stroke-based sketch as a graph with nodes representing the sampled points along input strokes and edges encoding the stroke structure information. To predict the per-node labels, our SketchGNN uses graph convolution and a static-dynamic branching network architecture to extract the features at three levels, i.e., point-level, stroke-level, and sketch-level. SketchGNN significantly improves the accuracy of the state-of-the-art methods for semantic sketch segmentation (by 11.2% in the pixel-based metric and 18.2% in the component-based metric over a large-scale challenging SPG dataset) and has magnitudes fewer parameters than both image-based and sequence-based methods.
TL;DR: In this article, a differentiable ray tracing image formation model is proposed to render optical images in the full field by taking into account all on/off-axis aberrations governed by the theory of geometric optics.
Abstract: Imaging systems have long been designed in separated steps: experience-driven optical design followed by sophisticated image processing. Although recent advances in computational imaging aim to bridge the gap in an end-to-end fashion, the image formation models used in these approaches have been quite simplistic, built either on simple wave optics models such as Fourier transform, or on similar paraxial models. Such models only support the optimization of a single lens surface, which limits the achievable image quality. To overcome these challenges, we propose a general end-to-end complex lens design framework enabled by a differentiable ray tracing image formation model. Specifically, our model relies on the differentiable ray tracing rendering engine to render optical images in the full field by taking into account all on/off-axis aberrations governed by the theory of geometric optics. Our design pipeline can jointly optimize the lens module and the image reconstruction network for a specific imaging task. We demonstrate the effectiveness of the proposed method on two typical applications, including large field-of-view imaging and extended depth-of-field imaging. Both simulation and experimental results show superior image quality compared with conventional lens designs. Our framework offers a competitive alternative for the design of modern imaging systems.
TL;DR: In this paper, a neural technique for articulating 3D characters using enveloping with a pre-defined skeletal structure is developed, which produces high quality pose-dependent deformations.
Abstract: Animating a newly designed character using motion capture (mocap) data is a long standing problem in computer animation. A key consideration is the skeletal structure that should correspond to the available mocap data, and the shape deformation in the joint regions, which often requires a tailored, pose-specific refinement. In this work, we develop a neural technique for articulating 3D characters using enveloping with a pre-defined skeletal structure which produces high quality pose dependent deformations. Our framework learns to rig and skin characters with the same articulation structure (e.g., bipeds or quadrupeds), and builds the desired skeleton hierarchy into the network architecture. Furthermore, we propose neural blend shapes - a set of corrective pose-dependent shapes which improve the deformation quality in the joint regions in order to address the notorious artifacts resulting from standard rigging and skinning. Our system estimates neural blend shapes for input meshes with arbitrary connectivity, as well as weighting coefficients which are conditioned on the input joint rotations. Unlike recent deep learning techniques which supervise the network with ground-truth rigging and skinning parameters, our approach does not assume that the training data has a specific underlying deformation model. Instead, during training, the network observes deformed shapes and learns to infer the corresponding rig, skin and blend shapes using indirect supervision. During inference, we demonstrate that our network generalizes to unseen characters with arbitrary mesh connectivity, including unrigged characters built by 3D artists. Conforming to standard skeletal animation models enables direct plug-and-play in standard animation software, as well as game engines.
TL;DR: NeuMIP as mentioned in this paper generalizes traditional mipmap pyramids to pyramids of neural textures, combined with a fully connected network, and introduces neural offsets, a novel method which enables rendering materials with intricate parallax effects without any tessellation.
Abstract: We propose NeuMIP, a neural method for representing and rendering a variety of material appearances at different scales. Classical prefiltering (mipmapping) methods work well on simple material properties such as diffuse color, but fail to generalize to normals, self-shadowing, fibers or more complex microstructures and reflectances. In this work, we generalize traditional mipmap pyramids to pyramids of neural textures, combined with a fully connected network. We also introduce neural offsets, a novel method which enables rendering materials with intricate parallax effects without any tessellation. This generalizes classical parallax mapping, but is trained without supervision by any explicit heightfield. Neural materials within our system support a 7-dimensional query, including position, incoming and outgoing direction, and the desired filter kernel size. The materials have small storage (on the order of standard mipmapping except with more texture channels), and can be integrated within common Monte-Carlo path tracing systems. We demonstrate our method on a variety of materials, resulting in complex appearance across levels of detail, with accurate parallax, self-shadowing, and other effects.
TL;DR: In this paper, a global-mesh-guided alternating optimization algorithm is proposed to extract a two-layer geometric representation for depth and reflection reconstruction, view selection for temporally coherent view-warping, and smooth rendering refinements.
Abstract: This paper proposes a novel scalable image-based rendering (IBR) pipeline for indoor scenes with reflections. We make substantial progress towards three sub-problems in IBR, namely, depth and reflection reconstruction, view selection for temporally coherent view-warping, and smooth rendering refinements. First, we introduce a global-mesh-guided alternating optimization algorithm that robustly extracts a two-layer geometric representation. The front and back layers encode the RGB-D reconstruction and the reflection reconstruction, respectively. This representation minimizes the image composition error under novel views, enabling accurate renderings of reflections. Second, we introduce a novel approach to select adjacent views and compute blending weights for smooth and temporal coherent renderings. The third contribution is a supersampling network with a motion vector rectification module that refines the rendering results to improve the final output's temporal coherence. These three contributions together lead to a novel system that produces highly realistic rendering results with various reflections. The rendering quality outperforms state-of-the-art IBR or neural rendering algorithms considerably.
TL;DR: In this paper, a differentiable pipeline for co-designing a soft swimmer's geometry and controller is presented, which unlocks gradientbased algorithms for discovering novel swimmer designs more efficiently than traditional gradient-free solutions.
Abstract: The computational design of soft underwater swimmers is challenging because of the high degrees of freedom in soft-body modeling. In this paper, we present a differentiable pipeline for co-designing a soft swimmer's geometry and controller. Our pipeline unlocks gradient-based algorithms for discovering novel swimmer designs more efficiently than traditional gradient-free solutions. We propose Wasserstein barycenters as a basis for the geometric design of soft underwater swimmers since it is differentiable and can naturally interpolate between bio-inspired base shapes via optimal transport. By combining this design space with differentiable simulation and control, we can efficiently optimize a soft underwater swimmer's performance with fewer simulations than baseline methods. We demonstrate the efficacy of our method on various design problems such as fast, stable, and energy-efficient swimming and demonstrate applicability to multi-objective design.
TL;DR: This paper proposed a method for portrait relighting and background replacement, which maintains high-frequency boundary details and accurately synthesizes the subject's appearance as lit by novel illumination, thereby producing realistic composite images for any desired scene.
Abstract: We propose a novel system for portrait relighting and background replacement, which maintains high-frequency boundary details and accurately synthesizes the subject's appearance as lit by novel illumination, thereby producing realistic composite images for any desired scene. Our technique includes foreground estimation via alpha matting, relighting, and compositing. We demonstrate that each of these stages can be tackled in a sequential pipeline without the use of priors (e.g. known background or known illumination) and with no specialized acquisition techniques, using only a single RGB portrait image and a novel, target HDR lighting environment as inputs. We train our model using relit portraits of subjects captured in a light stage computational illumination system, which records multiple lighting conditions, high quality geometry, and accurate alpha mattes. To perform realistic relighting for compositing, we introduce a novel per-pixel lighting representation in a deep learning framework, which explicitly models the diffuse and the specular components of appearance, producing relit portraits with convincingly rendered non-Lambertian effects like specular highlights. Multiple experiments and comparisons show the effectiveness of the proposed approach when applied to in-the-wild images.
TL;DR: Fusion 360 Gallery as discussed by the authors is a dataset of 8,625 human design sequences expressed in a simple language with just the sketch and extrude modeling operations, and an interactive environment called Fusion 360 Gym, which exposes the sequential construction of a CAD program as a Markov decision process, making it amendable to machine learning approaches.
Abstract: Parametric computer-aided design (CAD) is a standard paradigm used to design manufactured objects, where a 3D shape is represented as a program supported by the CAD software. Despite the pervasiveness of parametric CAD and a growing interest from the research community, currently there does not exist a dataset of realistic CAD models in a concise programmatic form. In this paper we present the Fusion 360 Gallery, consisting of a simple language with just the sketch and extrude modeling operations, and a dataset of 8,625 human design sequences expressed in this language. We also present an interactive environment called the Fusion 360 Gym, which exposes the sequential construction of a CAD program as a Markov decision process, making it amendable to machine learning approaches. As a use case for our dataset and environment, we define the CAD reconstruction task of recovering a CAD program from a target geometry. We report results of applying state-of-the-art methods of program synthesis with neurally guided search on this task.
TL;DR: In this paper, online reconstruction based on RGB-D sequences has thus far been restrained to relatively slow camera motions (e.g., slow camera motion) due to the slow camera movements.
Abstract: Online reconstruction based on RGB-D sequences has thus far been restrained to relatively slow camera motions (