Animation techniques bring digital 3D worlds and characters to life. However, manual animation is tedious and automated techniques are often specialized to narrow shape classes. In our work, we propose a technique for automatic re-animation of various 3D shapes based on a motion prior extracted from a video diffusion model. Unlike existing$4 D$generation methods, we focus solely on the motion, and we leverage an explicit mesh-based representation compatible with existing computer-graphics pipelines. Furthermore, our utilization of diffusion features enhances accuracy of our motion fitting. We analyze efficacy of these features for animation fitting and we experimentally validate our approach for two different diffusion models and four animation models. Finally, we demonstrate that our time-efficient zero-shot method achieves a superior performance re-animating a diverse set of 3D shapes when compared to existing techniques in a user study.

Many 3D tasks such as pose alignment, animation, motion transfer, and 3D reconstruction rely on establishing correspondences between 3D shapes. This challenge has recently been approached by pairwise matching of semantic features from pre-trained vision models. However, despite their power, these features struggle to differentiate instances of the same semantic class such as “left hand” versus “right hand” which leads to substantial mapping errors. To solve this, we learn a surface-aware embedding space that is robust to these ambiguities while facilitating shared mapping for an entire family of 3D shapes. Importantly, our approach is self-supervised and requires only a small number of unpaired training meshes to infer features for new possibly imperfect 3D shapes at test time. We achieve this by introducing a contrastive loss that preserves the semantic content of the features distilled from foundational models while disambiguating features located far apart on the shape’s surface. We observe superior performance in correspondence matching benchmarks and enable downstream applications including 2D-to-3D and 3D-to-3D texture transfer, in-part segmentation, pose alignment, and motion transfer in low-data regimes. Unlike previous pairwise approaches, our solution constructs a joint embedding space, where both seen and unseen 3D shapes are implicitly aligned without further optimization. The code is available at https://graphics.tudelft.nl/SurfaceAware3DFeatures.

Efficient and precise texture filtering is essential in various applications. However, there is often a trade-off between coarse real-time approximations and accurate computationally-expensive supersampling. We introduce a novel efficient texture-filtering method over arbitrary quadrilateral footprints, achieving high accuracy at a low computational cost. We achieve this by pre-computing integration tables that sparsely sample the space of possible footprints. Finally, we compare the qualitative and computational performance of our method to commonly used techniques and demonstrate various applications for high-quality real-time image synthesis, including normal filtering, soft shadow mapping, and glint rendering.

Dynamic Neural Radiance Fields (NeRFs) achieve remarkable visual quality when synthesizing novel views of time-evolving 3D scenes. However, the common reliance on backward deformation fields makes reanimation of the captured object poses challenging. Moreover, the state of the art dynamic models are often limited by low visual fidelity, long reconstruction time or specificity to narrow application domains.  In this paper, we present a novel method utilizing a point-based representation and Linear Blend Skinning (LBS) to jointly learn a Dynamic NeRF and an associated skeletal model from even sparse multi-view video. Our forward-warping approach achieves state-of-the-art visual fidelity when synthesizing novel views and poses while significantly reducing the necessary learning time when compared to existing work. We demonstrate the versatility of our representation on a variety of articulated objects from common datasets and obtain reposable 3D reconstructions without the need of object-specific skeletal templates.

Foveated graphics is a promising approach to solving the bandwidth challenges of immersive virtual and augmented reality displays by exploiting the falloff in spatial acuity in the periphery of the visual field. However, the perceptual models used in these applications neglect the effects of higherlevel cognitive processing, namely the allocation of visual attention, and are thus overestimating sensitivity in the periphery in many scenarios. Here, we introduce the first attention-aware model of contrast sensitivity. We conduct user studies to measure contrast sensitivity under different attention distributions and show that sensitivity in the periphery drops significantly when the user is required to allocate attention to the fovea. We motivate the development of future foveation models with another user study and demonstrate that tolerance for foveation in the periphery is significantly higher when the user is concentrating on a task in the fovea. Analysis of our model predicts significant bandwidth savings over those afforded by current models. As such, our work forms the foundation for attention-aware foveated graphics techniques.

Unsupervised learning of 3D-aware generative adversarial networks (GANs) using only collections of single-view 2D photographs has very recently made much progress. These 3D GANs, however, have not been demonstrated for human bodies and the generated radiance fields of existing frameworks are not directly editable, limiting their applicability in downstream tasks. We propose a solution to these challenges by developing a 3D GAN framework that learns to generate radiance fields of human bodies or faces in a canonical pose and warp them using an explicit deformation field into a desired body pose or facial expression. Using our framework, we demonstrate the first high-quality radiance field generation results for human bodies. Moreover, we show that our deformation-aware training procedure significantly improves the quality of generated bodies or faces when editing their poses or facial expressions compared to a 3D GAN that is not trained with explicit deformations.

We have witnessed rapid progress on 3D-aware image synthesis, leveraging recent advances in generative visual models and neural rendering. Existing approaches how-ever fall short in two ways: first, they may lack an under-lying 3D representation or rely on view-inconsistent rendering, hence synthesizing images that are not multi-view consistent; second, they often depend upon representation network architectures that are not expressive enough, and their results thus lack in image quality. We propose a novel generative model, named Periodic Implicit Generative Adversarial Networks (π-GAN or pi-GAN), for high-quality 3D-aware image synthesis. π-GAN leverages neural representations with periodic activation functions and volumetric rendering to represent scenes as view-consistent radiance fields. The proposed approach obtains state-of-the-art results for 3D-aware image synthesis with multiple real and synthetic datasets.

Novel view synthesis is a challenging and ill-posed inverse rendering problem. Neural rendering techniques have recently achieved photorealistic image quality for this task. State-of-the-art (SOTA) neural volume rendering approaches, however, are slow to train and require minutes of inference (i.e., rendering) time for high image resolutions. We adopt high-capacity neural scene representations with periodic activations for jointly optimizing an implicit surface and a radiance field of a scene supervised exclusively with posed 2D images. Our neural rendering pipeline accelerates SOTA neural volume rendering by about two orders of magnitude and our implicit surface representation is unique in allowing us to export a mesh with view-dependent texture information. Thus, like other implicit surface representations, ours is compatible with traditional graphics pipelines, enabling real-time rendering rates, while achieving unprecedented image quality compared to other surface methods. We assess the quality of our approach using existing datasets as well as high-quality 3D face data captured with a custom multi-camera rig.

Novel view synthesis is a long-standing problem in machine learning and computer vision. Significant progress has recently been made in developing neural scene representations and rendering techniques that synthesize photorealistic images from arbitrary views. These representations, however, are extremely slow to train and often also slow to render. Inspired by neural variants of image-based rendering, we develop a new neural rendering approach with the goal of quickly learning a high-quality representation which can also be rendered in real-time. Our approach, MetaNLR++, accomplishes this by using a unique combination of a neural shape representation and 2D CNN-based image feature extraction, aggregation, and re-projection. To push representation convergence times down to minutes, we leverage meta learning to learn neural shape and image feature priors which accelerate training. The optimized shape and image features can then be extracted using traditional graphics techniques and rendered in real time. We show that MetaNLR++ achieves similar or better novel view synthesis results in a fraction of the time that competing methods require.

Virtual and augmented reality (VR/AR) displays strive to provide a resolution, framerate and field of view that matches the perceptual capabilities of the human visual system, all while constrained by limited compute budgets and transmission bandwidths of wearable computing systems. Foveated graphics techniques have emerged that could achieve these goals by exploiting the falloff of spatial acuity in the periphery of the visual field. However, considerably less attention has been given to temporal aspects of human vision, which also vary across the retina. This is in part due to limitations of current eccentricity-dependent models of the visual system. We introduce a new model, experimentally measuring and computationally fitting eccentricity-dependent critical flicker fusion thresholds jointly for both space and time. In this way, our model is unique in enabling the prediction of temporal information that is imperceptible for a certain spatial frequency, eccentricity, and range of luminance levels. We validate our model with an image quality user study, and use it to predict potential bandwidth savings 7X higher than those afforded by current spatial-only foveated models. As such, this work forms the enabling foundation for new temporally foveated graphics techniques.

Virtual and augmented reality (VR/AR) displays crucially rely on stereoscopic rendering to enable perceptually realistic user experiences. Yet, existing near-eye display systems ignore the gaze-dependent shift of the no-parallax point in the human eye. Here, we introduce a gaze-contingent stereo rendering technique that models this effect and conduct several user studies to validate its effectiveness. Our findings include experimental validation of the location of the no-parallax point, which we then use to demonstrate significant improvements of disparity and shape distortion in a VR setting, and consistent alignment of physical and digitally rendered objects across depths in optical see-through AR. Our work shows that gaze-contingent stereo rendering improves perceptual realism and depth perception of emerging wearable computing systems.

Understanding where people are looking is an informative social cue. In this work, we present Gaze360, a large-scale remote gaze-tracking dataset and method for robust 3D gaze estimation in unconstrained images. Our dataset consists of 238 subjects in indoor and outdoor environments with labelled 3D gaze across a wide range of head poses and distances. It is the largest publicly available dataset of its kind by both subject and variety, made possible by a simple and efficient collection method. Our proposed 3D gaze model extends existing models to include temporal information and to directly output an estimate of gaze uncertainty. We demonstrate the benefits of our model via an ablation study, and show its generalization performance via a cross-dataset evaluation against other recent gaze benchmark datasets. We furthermore propose a simple self-supervised approach to improve cross-dataset domain adaptation. Finally, we demonstrate an application of our model for estimating customer attention in a supermarket setting. Our dataset and models will be made available at http://gaze360.csail.mit.edu.

Motivated by the recent potential of mass customization brought by whole-garment knitting machines, we introduce the new problem of automatic machine instruction generation using a single image of the desired physical product, which we apply to machine knitting. We propose to tackle this problem by directly learning to synthesize regular machine instructions from real images. We create a cured dataset of real samples with their instruction counterpart and propose to use synthetic images to augment it in a novel way. We theoretically motivate our data mixing framework and show empirical results suggesting that making real images look more synthetic is beneficial in our problem setup.

We propose a system to infer binocular disparity from a monocular video stream in real-time. Different from classic reconstruction of physical depth in computer vision, we compute perceptually plausible disparity, that is numerically inaccurate, but results in a very similar overall depth impression with plausible overall layout, sharp edges, fine details and agreement between luminance and disparity. We use several simple monocular cues to estimate disparity maps and confidence maps of low spatial and temporal resolution in real-time. These are complemented by spatially-varying, appearance-dependent and class-specific disparity prior maps, learned from example stereo images. Scene classification selects this prior at runtime. Fusion of prior and cues is done by means of robust MAP inference on a dense spatio-temporal conditional random field with high spatial and temporal resolution. Using normal distributions allows this in constant-time, parallel per-pixel work. We compare our approach to previous 2D-to-3D conversion systems in terms of different metrics, as well as a user study and validate our notion of perceptually plausible disparity.

Stereoscopic 3D (S3D) movies have become widely popular in the movie theaters, but the adoption of S3D at home is low even though most TV sets support S3D. It is widely believed that S3D with glasses is not the right approach for the home. A much more appealing approach is to use automulti-scopic displays that provide a glasses-free 3D experience to multiple viewers. A technical challenge is the lack of native multiview content that is required to deliver a proper view of the scene for every viewpoint. Our approach takes advantage of the abundance of stereoscopic 3D movies. We propose a real-time system that can convert stereoscopic video to a high-quality multiview video that can be directly fed to automultiscopic displays. Our algorithm uses a wavelet-based decomposition of stereoscopic images with per-wavelet disparity estimation. A key to our solution lies in combining Lagrangian and Eulerian approaches for both the disparity estimation and novel view synthesis, which leverages the complementary advantages of both techniques. The solution preserves all the features of Eulerian methods, e.g., subpixel accuracy, high performance, robustness to ambiguous depth cases, and easy integration of inter-view aliasing while maintaining the advantages of Lagrangian approaches, e.g., robustness to large disparities and possibility of performing non-trivial disparity manipulations through both view extrapolation and interpolation. The method achieves real-time performance on current GPUs. Its design also enables an easy hardware implementation that is demonstrated using a field-programmable gate array. We analyze the visual quality and robustness of our technique on a number of synthetic and real-world examples. We also perform a user experiment which demonstrates benefits of the technique when compared to existing solutions.

Accommodative depth cues, a wide field of view, and ever-higher resolutions all present major hardware design challenges for near-eye displays. Optimizing a design to overcome one of these challenges typically leads to a trade-off in the others. We tackle this problem by introducing an all-in-one solution - a new wide field of view, gaze-tracked near-eye display for augmented reality applications. The key component of our solution is the use of a single see-through, varifocal deformable membrane mirror for each eye reflecting a display. They are controlled by airtight cavities and change the effective focal power to present a virtual image at a target depth plane which is determined by the gaze tracker. The benefits of using the membranes include wide field of view (100° diagonal) and fast depth switching (from 20 cm to infinity within 300 ms). Our subjective experiment verifies the prototype and demonstrates its potential benefits for near-eye see-through displays.

From scientific research to commercial applications, eye tracking is an important tool across many domains. Despite its range of applications, eye tracking has yet to become a pervasive technology. We believe that we can put the power of eye tracking in everyone's palm by building eye tracking software that works on commodity hardware such as mobile phones and tablets, without the need for additional sensors or devices. We tackle this problem by introducing GazeCapture, the first large-scale dataset for eye tracking, containing data from over 1450 people consisting of almost 2:5M frames. Using GazeCapture, we train iTracker, a convolutional neural network for eye tracking, which achieves a significant reduction in error over previous approaches while running in real time (10-15fps) on a modern mobile device. Our model achieves a prediction error of 1.71cm and 2.53cm without calibration on mobile phones and tablets respectively. With calibration, this is reduced to 1.34cm and 2.12cm. Further, we demonstrate that the features learned by iTracker generalize well to other datasets, achieving state-of-the-art results. The code, data, and models are available at http://gazecapture.csail.mit.edu.

Binocular disparity is the main depth cue that makes stereoscopic images appear 3D. However, in many scenarios, the range of depth that can be reproduced by this cue is greatly limited and typically fixed due to constraints imposed by displays. For example, due to the low angular resolution of current automultiscopic screens, they can only reproduce a shallow depth range. In this work, we study the motion parallax cue, which is a relatively strong depth cue, and can be freely reproduced even on a 2D screen without any limits. We exploit the fact that in many practical scenarios, motion parallax provides sufficiently strong depth information that the presence of binocular depth cues can be reduced through aggressive disparity compression. To assess the strength of the effect we conduct psycho-visual experiments that measure the influence of motion parallax on depth perception and relate it to the depth resulting from binocular disparity. Based on the measurements, we propose a joint disparity-parallax computational model that predicts apparent depth resulting from both cues. We demonstrate how this model can be applied in the context of stereo and multiscopic image processing, and propose new disparity manipulation techniques, which first quantify depth obtained from motion parallax, and then adjust binocular disparity information accordingly. This allows us to manipulate the disparity signal according to the strength of motion parallax to improve the overall depth reproduction. This technique is validated in additional experiments.

Producing a high quality stereoscopic impression on current displays is a challenging task. The content has to be carefully prepared in order to maintain visual comfort, which typically affects the quality of depth reproduction. In this work, we show that this problem can be significantly alleviated when the eye fixation regions can be roughly estimated. We propose a new method for stereoscopic depth adjustment that utilizes eye tracking or other gaze prediction information. The key idea that distinguishes our approach from the previous work is to apply gradual depth adjustments at the eye fixation stage, so that they remain unnoticeable. To this end, we measure the limits imposed on the speed of disparity changes in various depth adjustment scenarios, and formulate a new model that can guide such seamless stereoscopic content processing. Based on this model, we propose a real-time controller that applies local manipulations to stereoscopic content to find the optimum between depth reproduction and visual comfort. We show that the controller is mostly immune to the limitations of low-cost eye tracking solutions. We also demonstrate benefits of our model in off-line applications, such as stereoscopic movie production, where skillful directors can reliably guide and predict viewers' attention or where attended image regions are identified during eye tracking sessions. We validate both our model and the controller in a series of user experiments. They show significant improvements in depth perception without sacrificing the visual quality when our techniques are applied.

Different from classic reconstruction of physical depth in computer vision, depth for 2D-to-3D stereo conversion is assigned by humans using semi-automatic painting interfaces and, consequently, is often dramatically wrong. Here we seek to better understand why it still does not fail to convey a sensation of depth. To this end, four typical disparity distortions resulting from manual 2D-to-3D stereo conversion are analyzed: i) smooth remapping, ii) spatial smoothness, iii) motion-compensated, temporal smoothness, and iv) completeness. A perceptual experiment is conducted to quantify the impact of each distortion on the plausibility of the 3D impression relative to a reference without distortion. Close-to-natural videos with known depth were distorted in one of the four above-mentioned aspects and subjects had to indicate if the distortion still allows for a plausible 3D effect. The smallest amounts of distortion that result in a significant rejection suggests a conservative upper bound on the quality requirement of 2D-to-3D conversion.

Abstract When human luminance perception operates close to its absolute threshold, i. e., the lowest perceivable absolute values, appearance changes substantially compared to common photopic or scotopic vision. In particular, most observers report perceiving temporally‐varying noise. Two reasons are physiologically plausible; quantum noise (due to the low absolute number of photons) and spontaneous photochemical reactions. Previously, static noise with a normal distribution and no account for absolute values was combined with blue hue shift and blur to simulate scotopic appearance on a photopic display for movies and interactive applications (e.g., games). We present a computational model to reproduce the specific distribution and dynamics of “scotopic noise” for specific absolute values. It automatically introduces a perceptually‐calibrated amount of noise for a specific luminance level and supports animated imagery. Our simulation runs in milliseconds at HD resolution using graphics hardware and favorably compares to simpler alternatives in a perceptual experiment.

Abstract Presenting stereoscopic content on 3D displays is a challenging task, usually requiring manual adjustments. A number of techniques have been developed to aid this process, but they account for binocular disparity of surfaces that are diffuse and opaque only. However, combinations of transparent as well as specular materials are common in the real and virtual worlds, and pose a significant problem. For example, excessive disparities can be created which cannot be fused by the observer. Also, multiple stereo interpretations become possible, e. g., for glass, that both reflects and refracts, which may confuse the observer and result in poor 3D experience. In this work, we propose an efficient method for analyzing and controlling disparities in computer‐generated images of such scenes where surface positions and a layer decomposition are available. Instead of assuming a single per‐pixel disparity value, we estimate all possibly perceived disparities at each image location. Based on this representation, we define an optimization to find the best per‐pixel camera parameters, assuring that all disparities can be easily fused by a human. A preliminary perceptual study indicates, that our approach combines comfortable viewing with realistic depiction of typical specular scenes.

Several approaches attempt to reproduce the appearance of a scotopic low-light night scene on a photopic display (“day-for-night”) by introducing color desaturation, loss of acuity, and the Purkinje shift toward blue colors. We argue that faithful stereo reproduction of night scenes on photopic stereo displays requires manipulation of not only color but also binocular disparity. To this end, we performed a psychophysics experiment to devise a model of disparity at scotopic luminance levels. Using this model, we can match binocular disparity of a scotopic stereo content displayed on a photopic monitor to the disparity that would be perceived if the scene was actually scotopic. The model allows for real-time computation of common stereo content as found in interactive applications such as simulators or computer games.

Abstract Beyond the careful design of stereo acquisition equipment and rendering algorithms, disparity post‐processing has recently received much attention, where one of the key tasks is to compress the originally large disparity range to avoid viewing discomfort. The perception of dynamic stereo content however, relies on reproducing the full disparity‐time volume that a scene point undergoes in motion. This volume can be strongly distorted in manipulation, which is only concerned with changing disparity at one instant in time, even if the temporal coherence of that change is maintained. We propose an optimization to preserve stereo motion of content that was subject to an arbitrary disparity manipulation, based on a perceptual model of temporal disparity changes. Furthermore, we introduce a novel 3D warping technique to create stereo image pairs that conform to this optimized disparity map. The paper concludes with perceptual studies of motion reproduction quality and task performance in a simple game, showing how our optimization can achieve both viewing comfort and faithful stereo motion.