SIGGRAPH 2017 论文列表

Increasing resolution and dynamic range of digital color displays is challenging with designs confined by cost and power specifications. This necessitates modern displays to trade-off spatial and temporal resolution for color reproduction capability. In this work we explore the idea of joint hardware and algorithm design to balance such trade-offs. We introduce a system that uses content-adaptive and compressive factorizations to reproduce colors. Each target frame is factorized into two products of high-resolution monochromatic and low-resolution color images, which then get integrated through temporal or spatial multiplexing. As our framework minimizes the error in colorimetric space, the perceived color rendition is high, and thanks to GPU acceleration, the results are generated in real-time. We evaluate our system with a LCD prototype that uses LED backlight array and temporal multiplexing to reproduce color images. Our approach enables high effective resolution and dynamic range without increasing power consumption. We also demonstrate low-cost extensions to hyperspectral and light-field imaging, which are possible due to compressive nature of our system.

A number of consumer-grade spherical cameras have recently appeared, enabling affordable monoscopic VR content creation in the form of full 360° X 180° spherical panoramic photos and videos. While monoscopic content is certainly engaging, it fails to leverage a main aspect of VR HMDs, namely stereoscopic display. Recent stereoscopic capture rigs involve placing many cameras in a ring and synthesizing an omni-directional stereo panorama enabling a user to look around to explore the scene in stereo. In this work, we describe a method that takes images from two 360° spherical cameras and synthesizes an omni-directional stereo panorama with stereo in all directions. Our proposed method has a lower equipment cost than camera-ring alternatives, can be assembled with currently available off-the-shelf equipment, and is relatively small and light-weight compared to the alternatives. We validate our method by generating both stills and videos. We have conducted a user study to better understand what kinds of geometric processing are necessary for a pleasant viewing experience. We also discuss several algorithmic variations, each with their own time and quality trade-offs.

Rendering explosions with self-illumination is a challenging problem. Explosions contain animated volumetric light sources immersed in animated smoke that cast volumetric shadows, which play an essential role and are expensive to compute. We propose an efficient solution that redefines this problem as rendering with many animated lights by converting the volumetric lighting data into a large number of point lights. Focusing on temporal coherency to avoid flickering in animations, we introduce lighting grid hierarchy for approximating the volumetric illumination at different resolutions. Using this structure we can efficiently approximate the lighting at any point inside or outside of the explosion volume as a mixture of lighting contributions from all levels of the hierarchy. As a result, we are able to capture high-frequency details of local illumination, as well as the potentially strong impact of distant illumination. Most importantly, this hierarchical structure allows us to efficiently precompute volumetric shadows, which substantially accelerates the lighting computation. Finally, we provide a scalable approach for computing the multiple scattering of light within the smoke volume using our lighting grid hierarchy. Temporal coherency is achieved by relying on continuous formulations at all stages of the lighting approximation. We show that our method is efficient and effective approximating the self-illumination of explosions with visually indistinguishable results, as compared to path tracing. We also show that our method can be applied to other problems involving a large number of (animated) point lights.

Given a robust control system, physical simulation offers the potential for interactive human characters that move in realistic and responsive ways. In this article, we describe how to learn a scheduling scheme that reorders short control fragments as necessary at runtime to create a control system that can respond to disturbances and allows steering and other user interactions. These schedulers provide robust control of a wide range of highly dynamic behaviors, including walking on a ball, balancing on a bongo board, skateboarding, running, push-recovery, and breakdancing. We show that moderate-sized Q-networks can model the schedulers for these control tasks effectively and that those schedulers can be efficiently learned by the deep Q-learning algorithm.

We propose a method for converting geometric shapes into hierarchically segmented parts with part labels. Our key idea is to train category-specific models from the scene graphs and part names that accompany 3D shapes in public repositories. These freely-available annotations represent an enormous, untapped source of information on geometry. However, because the models and corresponding scene graphs are created by a wide range of modelers with different levels of expertise, modeling tools, and objectives, these models have very inconsistent segmentations and hierarchies with sparse and noisy textual tags. Our method involves two analysis steps. First, we perform a joint optimization to simultaneously cluster and label parts in the database while also inferring a canonical tag dictionary and part hierarchy. We then use this labeled data to train a method for hierarchical segmentation and labeling of new 3D shapes. We demonstrate that our method can mine complex information, detecting hierarchies in man-made objects and their constituent parts, obtaining finer scale details than existing alternatives. We also show that, by performing domain transfer using a few supervised examples, our technique outperforms fully-supervised techniques that require hundreds of manually-labeled models.

Regression-based algorithms have shown to be good at denoising Monte Carlo (MC) renderings by leveraging its inexpensive by-products (e.g., feature buffers). However, when using higher-order models to handle complex cases, these techniques often overfit to noise in the input. For this reason, supervised learning methods have been proposed that train on a large collection of reference examples, but they use explicit filters that limit their denoising ability. To address these problems, we propose a novel, supervised learning approach that allows the filtering kernel to be more complex and general by leveraging a deep convolutional neural network (CNN) architecture. In one embodiment of our framework, the CNN directly predicts the final denoised pixel value as a highly non-linear combination of the input features. In a second approach, we introduce a novel, kernel-prediction network which uses the CNN to estimate the local weighting kernels used to compute each denoised pixel from its neighbors. We train and evaluate our networks on production data and observe improvements over state-of-the-art MC denoisers, showing that our methods generalize well to a variety of scenes. We conclude by analyzing various components of our architecture and identify areas of further research in deep learning for MC denoising.

We present a novel method to interpolate smoke and liquid simulations in order to perform data-driven fluid simulations. Our approach calculates a dense space-time deformation using grid-based signed-distance functions of the inputs. A key advantage of this implicit Eulerian representation is that it allows us to use powerful techniques from the optical flow area. We employ a five-dimensional optical flow solve. In combination with a projection algorithm, and residual iterations, we achieve a robust matching of the inputs. Once the match is computed, arbitrary in-between variants can be created very efficiently. To concatenate multiple long-range deformations, we propose a novel alignment technique. Our approach has numerous advantages, including automatic matches without user input, volumetric deformations that can be applied to details around the surface, and the inherent handling of topology changes. As a result, we can interpolate swirling smoke clouds, and splashing liquid simulations. We can even match and interpolate phenomena with fundamentally different physics: a drop of liquid, and a blob of heavy smoke.

We present an approach to generate plausible acoustic effects at interactive rates in large dynamic environments containing many sound sources. Our formulation combines listener-based backward ray tracing with sound source clustering and hybrid audio rendering to handle complex scenes. We present a new algorithm for dynamic late reverberation that performs high-order ray tracing from the listener against spherical sound sources. We achieve sublinear scaling with the number of sources by clustering distant sound sources and taking relative visibility into account. We also describe a hybrid convolution-based audio rendering technique that can process hundreds of thousands of sound paths at interactive rates. We demonstrate the performance on many indoor and outdoor scenes with up to 200 sound sources. In practice, our algorithm can compute more than 50 reflection orders at interactive rates on a multicore PC, and we observe a 5x speedup over prior geometric sound propagation algorithms.

We describe a machine learning technique for reconstructing image sequences rendered using Monte Carlo methods. Our primary focus is on reconstruction of global illumination with extremely low sampling budgets at interactive rates. Motivated by recent advances in image restoration with deep convolutional networks, we propose a variant of these networks better suited to the class of noise present in Monte Carlo rendering. We allow for much larger pixel neighborhoods to be taken into account, while also improving execution speed by an order of magnitude. Our primary contribution is the addition of recurrent connections to the network in order to drastically improve temporal stability for sequences of sparsely sampled input images. Our method also has the desirable property of automatically modeling relationships based on auxiliary per-pixel input channels, such as depth and normals. We show significantly higher quality results compared to existing methods that run at comparable speeds, and furthermore argue a clear path for making our method run at realtime rates in the near future.

Due to the widespread use of compositing in contemporary feature films, green-screen keying has become an essential part of postproduction workflows. To comply with the ever-increasing quality requirements of the industry, specialized compositing artists spend countless hours using multiple commercial software tools, while eventually having to resort to manual painting because of the many shortcomings of these tools. Due to the sheer amount of manual labor involved in the process, new green-screen keying approaches that produce better keying results with less user interaction are welcome additions to the compositing artist’s arsenal. We found that—contrary to the common belief in the research community—production-quality green-screen keying is still an unresolved problem with its unique challenges. In this article, we propose a novel green-screen keying method utilizing a new energy minimization-based color unmixing algorithm. We present comprehensive comparisons with commercial software packages and relevant methods in literature, which show that the quality of our results is superior to any other currently available green-screen keying solution. It is important to note that, using the proposed method, these high-quality results can be generated using only one-tenth of the manual editing time that a professional compositing artist requires to process the same content having all previous state-of-the-art tools at one’s disposal.

We present a computational approach to creating animated plushies, soft robotic plush toys specifically-designed to reenact user-authored motions. Our design process is inspired by muscular hydrostat structures, which drive highly versatile motions in many biological systems. We begin by instrumenting simulated plush toys with a large number of small, independently-actuated, virtual muscle-fibers. Through an intuitive posing interface, users then begin animating their plushie. A novel numerical solver, reminiscent of inverse-kinematics, computes optimal contractions for each muscle-fiber such that the soft body of the plushie deforms to best match user input. By analyzing the co-activation patterns of the fibers that contribute most to the plushie's motions, our design system generates physically-realizable winch-tendon networks. Winch-tendon networks model the motorized cable-driven actuation mechanisms that drive the motions of our real-life plush toy prototypes. We demonstrate the effectiveness of our computational approach by co-designing motions and actuation systems for a variety of physically-simulated and fabricated plushies.

High-quality hand-made furniture often employs intrinsic joints that geometrically interlock along mating surfaces. Such joints increase the structural integrity of the furniture and add to its visual appeal. We present an interactive tool for designing such intrinsic joints. Users draw the visual appearance of the joints on the surface of an input furniture model as groups of two-dimensional (2D) regions that must belong to the same part. Our tool automatically partitions the furniture model into a set of solid 3D parts that conform to the user-specified 2D regions and assemble into the furniture. If the input does not merit assemblable solid 3D parts, then our tool reports the failure and suggests options for redesigning the 2D surface regions so that they are assemblable. Similarly, if any parts in the resulting assembly are unstable, then our tool suggests where additional 2D regions should be drawn to better interlock the parts and improve stability. To perform this stability analysis, we introduce a novel variational static analysis method that addresses shortcomings of the equilibrium method for our task. Specifically, our method correctly detects sliding instabilities and reports the locations and directions of sliding and hinging failures. We show that our tool can be used to generate over 100 joints inspired by traditional woodworking and Japanese joinery. We also design and fabricate nine complete furniture assemblies that are stable and connected using only the intrinsic joints produced by our tool.

Clebsch maps encode velocity fields through functions. These functions contain valuable information about the velocity field. For example, closed integral curves of the associated vorticity field are level lines of the vorticity Clebsch map. This makes Clebsch maps useful for visualization and fluid dynamics analysis. Additionally they can be used in the context of simulations to enhance flows through the introduction of subgrid vorticity. In this paper we study spherical Clebsch maps, which are particularly attractive. Elucidating their geometric structure, we show that such maps can be found as minimizers of a non-linear Dirichlet energy. To illustrate our approach we use a number of benchmark problems and apply it to numerically given flow fields. Code and a video can be found in the ACM Digital Library.

In this paper we introduce a novel method to adaptive incompressible SPH simulations. Instead of using a scheme with a number of fixed particle sizes or levels, our approach allows continuous particle sizes. This enables us to define optimal particle masses with respect to, e.g., the distance to the fluid's surface. A required change in mass due to the dynamics of the fluid is properly and stably handled by our scheme of mass redistribution. This includes temporally smooth changes in particle masses as well as sudden mass variations in regions of high flow dynamics. Our approach guarantees low spatial variations in particle size, which is a core property in order to achieve large adaptivity ratios for incompressible fluid simulations. Conceptually, our approach allows for infinite continuous adaptivity, practically we achieved adaptivity ratios up to 5 orders of magnitude, while still being mass preserving and numerically stable, yielding unprecedented vivid surface detail at comparably low computational cost and moderate particle counts.

This article presents a new version of the Gilbert-Johnson-Keerthi (GJK) algorithm that circumvents the shortcomings introduced by degenerate geometries. The original Johnson algorithm and Backup procedure are replaced by a distance subalgorithm that is faster and accurate to machine precision, thus guiding the GJK algorithm toward a shorter search path in less computing time. Numerical tests demonstrate that this effectively is a more robust procedure. In particular, when the objects are found in contact, the newly proposed subalgorithm runs from 15% to 30% times faster than the original one. The improved performance has a significant impact on various applications, such as real-time simulations and collision avoidance systems. Altogether, the main contributions made to the GJK algorithm are faster convergence rate and reduced computational time. These improvements may be easily added into existing implementations; furthermore, engineering applications that require solutions of distance queries to machine precision can now be tackled using the GJK algorithm.