In computed tomography (CT), the forward model consists of a linear Radon transform followed by an exponential nonlinearity based on the attenuation of light according to the Beer–Lambert Law. Conventional reconstruction often involves inverting this nonlinearity and then solving a linear inverse problem. However, this nonlinear measurement preprocessing is poorly conditioned in the vicinity of high-density materials, such as metal. This preprocessing makes CT reconstruction methods numerically sensitive and susceptible to artifacts near high-density regions. In this paper, we study a technique where the signal is directly reconstructed from raw measurements through the nonlinear forward model. Though this optimization is nonconvex, we show that gradient descent provably converges to the global optimum at a geometric rate, perfectly reconstructing the underlying signal with a near minimal number of random measurements. We also prove similar results in the under-determined setting where the number of measurements is significantly smaller than the dimension of the signal. This is achieved by enforcing prior structural information about the signal through constraints on the optimization variables. We illustrate the benefits of direct nonlinear CT reconstruction with cone-beam CT experiments on synthetic and real 3D volumes, in which metal artifacts are reduced compared to standard linear reconstruction methods. Our experiments also demonstrate that logarithmic preprocessing alone is sufficient to produce metal artifacts, even in the absence of other causes such as beam hardening.

Text-to-image diffusion models produce impressive results but are frustrating tools for artists who desire fine-grained control. For example, a common use case is to create images of a specific concept in novel contexts, i.e., "identity-preserving generation". This setting, along with many other tasks (e.g., relighting), is a natural fit for image+text-conditional generative models. However, there is insufficient high-quality paired data to train such a model directly. We propose Diffusion Self-Distillation, a method for using a pre-trained text-to-image model to generate its own dataset for text-conditioned image-to-image tasks. We first leverage a text-to-image diffusion model’s in-context generation ability to create grids of images and curate a large paired dataset with the help of a vision-language model. We then fine-tune the text-to-image model into a text+image-to-image model using the curated paired dataset. We demonstrate that Diffusion Self-Distillation outperforms existing zero-shot methods and is competitive with per-instance tuning techniques on a wide range of identity-preserving generation tasks, without requiring test-time optimization. Project page: primecai.github.io/dsd.

We introduce X-Dyna, a novel zero-shot, diffusion-based pipeline for animating a single human image using facial expressions and body movements derived from a driving video, that generates realistic, context-aware dynamics for both the subject and the surrounding environment. Building on prior approaches centered on human pose control, X-Dyna addresses key shortcomings causing the loss of dynamic details, enhancing the lifelike qualities of human video animations. At the core of our approach is the Dynamics-Adapter, a lightweight module that effectively integrates reference appearance context into the spatial attentions of the diffusion backbone while preserving the capacity of motion modules in synthesizing fluid and intricate dynamic details. Beyond body pose control, we connect a local control module with our model to capture identity-disentangled facial expressions, facilitating accurate expression transfer for enhanced realism in animated scenes. Together, these components form a unified framework capable of learning physical human motion and natural scene dynamics from a diverse blend of human and scene videos. Comprehensive qualitative and quantitative evaluations demonstrate that X-Dyna outperforms state-of- the-art methods, creating highly lifelike and expressive animations. The code is available at https://github.com/bytedance/X-Dyna

3D Gaussian Splatting (3DGS) has emerged as the state-of-the-art 3D reconstruction technique, offering high-quality results with fast training and rendering. However, its expressivity is limited as pixels covered by the same Gaussian share identical colors aside from a Gaussian falloff scaling factor, and individual Gaussians can only represent simple ellipsoids geometrically. To overcome these limitations, we integrate texture and alpha mapping from traditional graphics with 3DGS. Our approach augments each Gaussian with alpha, RGB, or RGBA texture maps to model spatially varying color and opacity across each Gaussian’s extent. This allows Gaussians to represent richer texture patterns and geometric structures beyond single-color ellipsoids. Notably, alpha-only texture maps significantly improve Gaussian expressivity, while further augmenting with RGB texture maps achieve maximum expressivity. We validate our method on a wide variety of standard benchmark datasets and our own custom captures at both the object and scene levels, and demonstrate image quality improvements over existing methods while using a similar or lower number of Gaussians.

Human communication is inherently multimodal, involving a combination of verbal and non-verbal cues such as speech, facial expressions, and body gestures. Modeling these behaviors is essential for understanding human interaction and for creating virtual characters that can communicate naturally in applications like games, films, and virtual reality. However, existing motion generation models are typically limited to specific input modalities-either speech, text, or motion data-and cannot fully leverage the diversity of available data. In this paper, we propose a novel framework that unifies verbal and non-verbal language using multimodal language models for human motion understanding and generation. This model is flexible in taking text, speech, and motion or any combination of them as input. Coupled with our novel pre-training strategy, our model not only achieves state-of-the-art performance on co-speech gesture generation but also requires much less data for training. Our model also unlocks an array of novel tasks such as editable gesture generation and emotion prediction from motion. We believe unifying the verbal and non-verbal language of human motion is essential for real-world applications, and language models offer a powerful approach to achieving this goal. Project page: languageofmotion.github.io.

We present Buffer Anytime, a framework for estimation of depth and normal maps (which we call geometric buffers) from video that eliminates the need for paired video–depth and video–normal training data. Instead of relying on large-scale annotated video datasets, we demonstrate high-quality video buffer estimation by leveraging single-image priors with temporal consistency constraints. Our zero-shot training strategy combines state-of-the-art image estimation models based on optical flow smoothness through a hybrid loss function, implemented via a lightweight temporal attention architecture. Applied to leading image models like Depth Anything V2 and Marigold-E2E-FT, our approach significantly improves temporal consistency while maintaining accuracy. Experiments show that our method not only outperforms image-based approaches but also achieves results comparable to state-of-the-art video models trained on large-scale paired video datasets, despite using no such paired video data.

Apparel is essential to human life, offering protection, mirroring cultural identities, and showcasing personal style. Yet, the creation of garments remains a time-consuming process, largely due to the manual work involved in designing them. To simplify this process, we introduce AIpparel, a multimodal foundation model for generating and editing sewing patterns. Our model fine-tunes state-of-the-art large multimodal models (LMMs) on a custom-curated large-scale dataset of over 120,000 unique garments, each with multimodal annotations including text, images, and sewing patterns. Additionally, we propose a novel tokenization scheme that concisely encodes these complex sewing patterns so that LLMs can learn to predict them efficiently. AIpparel achieves state-of-the-art performance in single-modal tasks, including text-to-garment and image-to-garment prediction, and enables novel multimodal garment generation applications such as interactive garment editing. The project website is at https: //georgenakayama.github.io/AIpparel/.

We present FLARE, a feed-forward model designed to infer high-quality camera poses and 3D geometry from uncalibrated sparse-view images (i.e., as few as 2-8 inputs), which is a challenging yet practical setting in real-world applications. Our solution features a cascaded learning paradigm with camera pose serving as the critical bridge, recognizing its essential role in mapping 3D structures onto 2D image planes. Concretely, FLARE starts with camera pose estimation, whose results condition the subsequent learning of geometric structure and appearance, optimized through the objectives of geometry reconstruction and novel-view synthesis. Utilizing large-scale public datasets for training, our method delivers state-of-the-art performance in the tasks of pose estimation, geometry reconstruction, and novel view synthesis, while maintaining the inference efficiency (i.e., less than 0.5 seconds). The project page and code can be found at: https://zhanghe3z.github.io/FLARE/

Vision-language-action models (VLAs) have shown potential in leveraging pretrained vision-language models and diverse robot demonstrations for learning generalizable sensorimotor control. While this paradigm effectively utilizes large-scale data from both robotic and non-robotic sources, current VLAs primarily focus on direct input–output mappings, lacking the intermediate reasoning steps crucial for complex manipulation tasks. As a result, existing VLAs lack temporal planning or reasoning capabilities. In this paper, we introduce a method that incorporates explicit visual chain-of-thought (CoT) reasoning into vision-language-action models (VLAs) by predicting future image frames autoregressively as visual goals before generating a short action sequence to achieve these goals. We introduce CoT-VLA, a state-of-the-art 7B VLA that can understand and generate visual and action tokens. Our experimental results demonstrate that CoT-VLA achieves strong performance, outperforming the state-of-the-art VLA model by 17% in real-world manipulation tasks and 6% in simulation benchmarks. Videos are available at: https://cot-vla.github.io/.

We present GroomLight, a novel method for relightable hair appearance modeling from multi-view images. Existing hair capture methods struggle to balance photorealistic rendering with relighting capabilities. Analytical material models, while physically grounded, often fail to fully capture appearance details. Conversely, neural rendering approaches excel at view synthesis but generalize poorly to novel lighting conditions. GroomLight addresses this challenge by combining the strengths of both paradigms. It employs an extended hair BSDF model to capture primary light transport and a light-aware residual model to reconstruct the remaining details. We further propose a hybrid inverse rendering pipeline to optimize both components, enabling high-fidelity relighting, view synthesis, and material editing. Extensive evaluations on real-world hair data demonstrate state-of-the-art performance of our method. Our project website is at: https://syntec-research.github.io/GroomLight.

In dynamic 3D environments, accurately updating scene representations over time is crucial for applications in robotics, mixed reality, and embodied AI. As scenes evolve, efficient methods to incorporate changes are needed to maintain up-to-date, high-quality reconstructions without the computational overhead of re-optimizing the entire scene. This paper introduces CL-Splats, which incrementally updates Gaussian splatting-based 3D representations from sparse scene captures. CL-Splats integrates a robust change-detection module that segments updated and static components within the scene, enabling focused, local optimization that avoids unnecessary re-computation. Moreover, CL-Splats supports storing and recovering previous scene states, facilitating temporal segmentation and new scene-analysis applications. Our extensive experiments demonstrate that CL-Splats achieves efficient updates with improved reconstruction quality over the state-of-the-art. This establishes a robust foundation for future real-time adaptation in 3D scene reconstruction tasks.

In this paper, we present a self-calibrating framework that jointly optimizes camera parameters, lens distortion and 3D Gaussian representations, enabling accurate and efficient scene reconstruction. In particular, our technique enables high-quality scene reconstruction from Large field-of-view (FOV) imagery taken with wide-angle lenses, allowing the scene to be modeled from a smaller number of images. Our approach introduces a novel method for modeling complex lens distortions using a hybrid network that combines invertible residual networks with explicit grids. This design effectively regularizes the optimization process, achieving greater accuracy than conventional camera models. Additionally, we propose a cubemap-based resampling strategy to support large FOV images without sacrificing resolution or introducing distortion artifacts. Our method is compatible with the fast rasterization of Gaussian Splatting, adaptable to a wide variety of camera lens distortion, and demonstrates state-of-the-art performance on both synthetic and real-world datasets.

This paper introduces CameraCtrl II, a framework that enables large-scale dynamic scene exploration through a camera-controlled video diffusion model. Previous camera-conditioned video generative models suffer from diminished video dynamics and limited range of viewpoints when generating videos with large camera movement. We take an approach that progressively expands the generation of dynamic scenes -- first enhancing dynamic content within individual video clip, then extending this capability to create seamless explorations across broad viewpoint ranges. Specifically, we construct a dataset featuring a large degree of dynamics with camera parameter annotations for training while designing a lightweight camera injection module and training scheme to preserve dynamics of the pretrained models. Building on these improved single-clip techniques, we enable extended scene exploration by allowing users to iteratively specify camera trajectories for generating coherent video sequences. Experiments across diverse scenarios demonstrate that CameraCtrl Ii enables camera-controlled dynamic scene synthesis with substantially wider spatial exploration than previous approaches.

Video diffusion models have recently shown promise for world modeling through autoregressive frame prediction conditioned on actions. However, they struggle to maintain long-term memory due to the high computational cost associated with processing extended sequences in attention layers. To overcome this limitation, we propose a novel architecture leveraging state-space models (SSMs) to extend temporal memory without compromising computational efficiency. Unlike previous approaches that retrofit SSMs for non-causal vision tasks, our method fully exploits the inherent advantages of SSMs in causal sequence modeling. Central to our design is a block-wise SSM scanning scheme, which strategically trades off spatial consistency for extended temporal memory, combined with dense local attention to ensure coherence between consecutive frames. We evaluate the long-term memory capabilities of our model through spatial retrieval and reasoning tasks over extended horizons. Experiments on Memory Maze and Minecraft datasets demonstrate that our approach surpasses baselines in preserving long-range memory, while maintaining practical inference speeds suitable for interactive applications.

Controllability plays a crucial role in video generation, as it allows users to create and edit content more precisely. Existing models, however, lack control of camera pose that serves as a cinematic language to express deeper narrative nuances. To alleviate this issue, we introduce \method, enabling accurate camera pose control for video diffusion models. Our approach explores effective camera trajectory parameterization along with a plug-and-play camera pose control module that is trained on top of a video diffusion model, leaving other modules of the base model untouched. Moreover, a comprehensive study on the effect of various training datasets is conducted, suggesting that videos with diverse camera distributions and similar appearance to the base model indeed enhance controllability and generalization. Experimental results demonstrate the effectiveness of \method in achieving precise camera control with different video generation models, marking a step forward in the pursuit of dynamic and customized video storytelling from textual and camera pose inputs.

Scene image editing is crucial for entertainment, photography, and advertising design. Existing methods solely focus on either 2D individual object or 3D global scene editing. This results in a lack of a unified approach to effectively control and manipulate scenes at the 3D level with different levels of granularity. In this work, we propose 3DitScene, a novel and unified scene editing framework leveraging language-guided disentangled Gaussian Splatting that enables seamless editing from 2D to 3D, allowing precise control over scene composition and individual objects. We first incorporate 3D Gaussians that are refined through generative priors and optimization techniques. Language features from CLIP then introduce semantics into 3D geometry for object disentanglement. With the disentangled Gaussians, 3DitScene allows for manipulation at both the global and individual levels, revolutionizing creative expression and empowering control over scenes and objects. Experimental results demonstrate the effectiveness and versatility of 3DitScene in scene image editing.

Diffusion models approximate the denoising distribution as a Gaussian and predict its mean, whereas flow matching models reparameterize the Gaussian mean as flow velocity. However, they underperform in few-step sampling due to discretization error and tend to produce over-saturated colors under classifier-free guidance (CFG). To address these limitations, we propose a novel Gaussian mixture flow matching (GMFlow) model: instead of predicting the mean, GMFlow predicts dynamic Gaussian mixture (GM) parameters to capture a multi-modal flow velocity distribution, which can be learned with a KL divergence loss. We demonstrate that GMFlow generalizes previous diffusion and flow matching models where a single Gaussian is learned with an $L_2$ denoising loss. For inference, we derive GM-SDE/ODE solvers that leverage analytic denoising distributions and velocity fields for precise few-step sampling. Furthermore, we introduce a novel probabilistic guidance scheme that mitigates the over-saturation issues of CFG and improves image generation quality. Extensive experiments demonstrate that GMFlow consistently outperforms flow matching baselines in generation quality, achieving a Precision of 0.942 with only 6 sampling steps on ImageNet 256$\times$256.

Reasoning stands as a cornerstone of intelligence, enabling the synthesis of existing knowledge to solve complex problems. Despite remarkable progress, existing reasoning benchmarks often fail to rigorously evaluate the nuanced reasoning capabilities required for complex, real-world problemsolving, particularly in multi-disciplinary and multimodal contexts. In this paper, we introduce a graduate-level, multi-disciplinary, EnglishChinese benchmark, dubbed as Reasoning Bench (RBench), for assessing the reasoning capability of both language and multimodal models. RBench spans 1,094 questions across 108 subjects for language model evaluation and 665 questions across 83 subjects for multimodal model testing. These questions are meticulously curated to ensure rigorous difficulty calibration, subject balance, and cross-linguistic alignment, enabling the assessment to be an Olympiad-level multidisciplinary benchmark. We evaluate many models such as o1, GPT-4o, DeepSeek-R1, etc. Experimental results indicate that advanced models perform poorly on complex reasoning, especially multimodal reasoning. Even the top-performing model OpenAI o1 achieves only 53.2% accuracy on our multimodal evaluation. Data and code are made publicly available athttps://evalmodels.github.io/rbench/

Volume parameterizations abound in recent literature, encompassing methods from classic voxel grids to implicit neural representations. While implicit representations offer impressive capacity and improved memory efficiency compared to voxel grids, they traditionally require training through nonconvex optimization, which can be slow and sensitive to initialization and hyperparameters. We introduce GA-Planes, a novel family of implicit neural volume representations inspired by Geometric Algebra that can be trained using convex optimization, addressing the limitations of nonconvex methods. GA-Planes models generalize many existing representations including any combination of features stored in tensor basis elements followed by a neural feature decoder, and can be adapted to convex or nonconvex training as needed for various inverse problems. In the 2D setting, we prove GA-Planes models are equivalent to a low-rank plus low-resolution matrix factorization that outperforms the classic low-rank plus sparse decomposition for fitting a natural image. In 3D, GA-Planes models exhibit competitive expressiveness, model size, and optimizability across tasks such as radiance field reconstruction, 3D segmentation, and video segmentation.

Protein structure prediction models are now capable of generating accurate 3D structural hypotheses from sequence alone. However, they routinely fail to capture the conformational diversity of dynamic biomolecular complexes, often requiring heuristic MSA subsampling approaches for generating alternative states. In parallel, cryo-electron microscopy (cryo-EM) has emerged as a powerful tool for imaging near-native structural heterogeneity, but is challenged by arduous pipelines to transform raw experimental data into atomic models. Here, we bridge the gap between these modalities, combining cryo-EM density maps with the rich sequence and biophysical priors learned by protein structure prediction models. Our method, CryoBoltz, guides the sampling trajectory of a pretrained biomolecular structure prediction model using both global and local structural constraints derived from density maps, driving predictions towards conformational states consistent with the experimental data. We demonstrate that this flexible yet powerful inference-time approach allows us to build atomic models into heterogeneous cryo-EM maps across a variety of dynamic biomolecular systems including transporters and antibodies.

Emerging world models autoregressively generate video frames in response to actions, such as camera movements and text prompts, among other control signals. Due to limited temporal context window sizes, these models often struggle to maintain scene consistency during revisits, leading to severe forgetting of previously generated environments. Inspired by the mechanisms of human memory, we introduce a novel framework to enhancing long-term consistency of video world models through a geometry-grounded long-term spatial memory. Our framework includes mechanisms to store and retrieve information from the long-term spatial memory and we curate custom datasets to train and evaluate world models with explicitly stored 3D memory mechanisms. Our evaluations show improved quality, consistency, and context length compared to relevant baselines, paving the way towards long-term consistent world generation.

We present a neural network structure, FramePack, to train next-frame (or next-frame-section) prediction models for video generation. FramePack compresses input frame contexts with frame-wise importance so that more frames can be encoded within a fixed context length, with more important frames having longer contexts. The frame importance can be measured using time proximity, feature similarity, or hybrid metrics. The packing method allows for inference with thousands of frames and training with relatively large batch sizes. We also present drift prevention methods to address observation bias (error accumulation), including early-established endpoints, adjusted sampling orders, and discrete history representation. Ablation studies validate the effectiveness of the anti-drifting methods in both single-directional video streaming and bi-directional video generation. Finally, we show that existing video diffusion models can be finetuned with FramePack, and analyze the differences between different packing schedules.

State-of-the-art neural rendering methods optimize Gaussian scene representations from a few photographs for novel-view synthesis. Building on these representations, we develop an efficient algorithm, dubbed Gaussian Wave Splatting, to turn these Gaussians into holograms. Unlike existing computergenerated holography (CGH) algorithms, Gaussian Wave Splatting supports accurate occlusions and view-dependent effects for photorealistic scenes by leveraging recent advances in neural rendering. Specifically, we derive a closed-form solution for a 2D Gaussian-to-hologram transform that supports occlusions and alpha blending. Inspired by classic computer graphics techniques, we also derive an efficient approximation of the aforementioned process in the Fourier domain that is easily parallelizable and implement it using custom CUDA kernels. By integrating emerging neural rendering pipelines with holographic display technology, our Gaussian-based CGH framework paves the way for next-generation holographic displays.

Generating photorealistic videos of digital humans in a controllable manner is crucial for a plethora of applications. Existing approaches either build on methods that employ template-based 3D representations or emerging video generation models but suffer from poor quality or limited consistency and identity preservation when generating individual or multiple digital humans. In this paper, we introduce a new interspatial attention (ISA) mechanism as a scalable building block for modern diffusion transformer (DiT)-based video generation models. ISA is a new type of cross attention that uses relative positional encodings tailored for the generation of human videos. Leveraging a custom-developed video variation autoencoder, we train a latent ISA-based diffusion model on a large corpus of video data. Our model achieves state-of-the-art performance for 4D human video synthesis, demonstrating remarkable motion consistency and identity preservation while providing precise control of the camera and body poses. Our code and model are publicly released at https://dsaurus.github.io/isa4d/.

Neural radiance fields (NeRFs) have demonstrated state-of-the-art performance for 3D computer vision tasks, including novel view synthesis and 3D shape reconstruction. However, these methods fail scattering medium, such as haze, is prevalent in the scene. To address this challenge, we introduce DehazeNeRF as a framework that robustly operates in hazy conditions. DehazeNeRF extends the volume rendering equation by adding physically realistic terms that model atmospheric scattering. By parameterizing these terms using suitable networks that match the physical properties, we introduce effective inductive biases, which, together with the proposed regularizations, allow DehazeNeRF to demonstrate successful multi-view haze removal, novel view synthesis, and 3D shape reconstruction where existing approaches failed. Our code and pretrained models can be found on this page.

Unsupervised learning of 3D human faces from unstructured 2D image data is an active research area. While recent works have achieved an impressive level of photorealism, they commonly lack control of lighting, which prevents the generated assets from being deployed in novel environments. To this end, we introduce LumiGAN, an unconditional Generative Adversarial Network (GAN) for 3D human faces with a physically based lighting module that enables relighting under novel illumination at inference time. Unlike prior work, LumiGAN can create realistic shadow effects using an efficient visibility formulation that is learned in a self-supervised manner. LumiGAN generates plausible physical properties for relightable faces, including surface normals, diffuse albedo, and specular tint without any ground truth data. In addition to state-of-the-art photorealism, we demonstrate significantly improved geometry generation compared to top-performing non-relightable 3D GANs and notably better relighting quality than existing relightable GANs.

Neural radiance fields enable novel-view synthesis and scene reconstruction with photorealistic quality from a few images, but require known and accurate camera poses. Conventional pose estimation algorithms fail on smooth or self-similar scenes, while methods performing inverse rendering from unposed views require a rough initialization of the camera orientations. The main difficulty of pose estimation lies in real-life objects being almost invariant under certain transformations, making the photometric distance between rendered views non-convex with respect to the camera parameters. Using an equivalence relation that matches the distribution of local minima in camera space, we reduce this space to its quotient set, in which gradient descent is more likely to converge. Using a neural network to regularize pose estimation, we demonstrate that our method – MELON – can reconstruct a neural radiance field from unposed images with state-of-the-art accuracy while requiring ten times fewer views than adversarial approaches. https://melon-nerf.github.io

Designing complex 3D scenes has been a tedious, manual process requiring domain expertise. Emerging text-to-3D generative models show great promise for making this task more intuitive, but existing approaches are limited to object-level generation. We introduce locally conditioned diffusion as an approach to compositional scene diffusion, providing control over semantic parts using text prompts and bounding boxes while ensuring seamless transitions between these parts. We demonstrate a score distillation sampling-based text-to-3D synthesis pipeline that enables compositional 3D scene generation at a higher fidelity than relevant baselines.

Access to high-quality and diverse 3D articulated digital human assets is crucial in various applications, ranging from virtual reality to social platforms. Generative approaches, such as 3D generative adversarial networks (GANs), are rapidly replacing laborious manual content creation tools. Existing 3D GAN frameworks typically rely on explicit representations that are fast to render but offer limited quality, or implicit representations, which offer high capacity but are slow to render, thereby limiting the 3D fidelity in GAN settings. In this work, we introduce layered surface volumes (LSVs) as a new 3D representation for articulatable humans. LSVs represent a human body using multiple textured mesh layers around a conventional mesh template. The explicit mesh layers can be interpreted as a discretized volume with finite thickness located around the template manifold, as such it has the capacity to capture off-template details like hair or accessories, while at the same time it can profit from mesh rasterization for fast rendering. LSVs can be articulated, and they exhibit exceptional efficiency in GAN settings. Trained on unstructured, single-view 2D image datasets, our LSV-GAN generates high-quality and viewconsistent 3D articulated digital humans without the need for view-inconsistent 2D upsampling networks. Project page can be found here.

Efficient generation of 3D digital humans is important in several industries, including virtual reality, social media, and cinematic production. 3D generative adversarial net-works (GANs) have demonstrated state-of-the-art (SOTA) quality and diversity for generated assets. Current 3D GAN architectures, however, typically rely on volume representations, which are slow to render, thereby hampering the GAN training and requiring multi- view-inconsistent 2D upsam-plers. Here, we introduce Gaussian Shell Maps (GSMs) as a framework that connects SOTA generator network archi-tectures with emerging 3D Gaussian rendering primitives using an articulable multi shell-based scaffold. In this set-ting, a CNN generates a 3D texture stack with features that are mapped to the shells. The latter represent inflated and deflated versions of a template surface of a digital human in a canonical body pose. Instead of rasterizing the shells directly, we sample 3D Gaussians on the shells whose at-tributes are encoded in the texture features. These Gaus-sians are efficiently and differentiably rendered. The ability to articulate the shells is important during GAN training and, at inference time, to deform a body into arbitrary user-defined poses. Our efficient rendering scheme bypasses the need for view-inconsistent upsamplers and achieves high-quality multi-view consistent renderings at a native resolution of 512 ×512 pixels. We demonstrate that GSMs suc-cessfully generate 3D humans when trained on single-view datasets, including SHHQ and DeepFashion. Project Page: rameenabdal.github.io/GaussianShellMaps

Recent breakthroughs in text-to-4D generation rely on pre-trained text-to-image and text-to-video models to generate dynamic 3D scenes. However, current text-to-4D methods face a three-way tradeoff between the quality of scene appearance, 3D structure, and motion. For example, text-to-image models and their 3D-aware variants are trained on internet-scale image datasets and can be used to produce scenes with realistic appearance and 3D structure—but no motion. Text-to-video models are trained on relatively smaller video datasets and can produce scenes with motion, but poorer appearance and 3D structure. While these models have complementary strengths, they also have opposing weaknesses, making it difficult to combine them in a way that alleviates this three-way tradeoff. Here, we introduce hybrid score distillation sampling, an alternating optimization procedure that blends supervision signals from multiple pre-trained diffusion models and incorporates benefits of each for high-fidelity text-to-4D generation. Using hybrid SDS, we demonstrate synthesis of 4D scenes with compelling appearance, 3D structure, and motion.

Traditional 3D content creation tools empower users to bring their imagination to life by giving them direct control over a scene's geometry, appearance, motion, and camera path. Creating computer-generated videos, however, is a tedious manual process, which can be automated by emerging text-to-video diffusion models. Despite great promise, video diffusion models are difficult to control, hindering a user to apply their own creativity rather than amplifying it. To address this challenge, we present a novel approach that combines the controllability of dynamic 3D meshes with the expressivity and editability of emerging diffusion models. For this purpose, our approach takes an animated, low-fidelity rendered mesh as input and injects the ground truth cor-respondence information obtained from the dynamic mesh into various stages of a pre-trained text-to-image generation model to output high-quality and temporally consistent frames. We demonstrate our approach on various examples where motion can be obtained by animating rigged assets or changing the camera path. Project page: primecai.github.io/ generative. rendering.

Customization techniques for text-to-image models have paved the way for a wide range of previously unattainable applications, enabling the generation of specific concepts across diverse contexts and styles. While existing methods facilitate high-fidelity customization for individual concepts or a limited, pre-defined set of them, they fall short of achieving scalability, where a single model can seamlessly render countless concepts. In this paper, we address a new problem called Modular Customization, with the goal of efficiently merging customized models that were fine-tuned independently for individual concepts. This allows the merged model to Jointly synthesize concepts in one image without compromising fidelity or incurring any additional computational costs. To address this problem, we introduce Orthogonal Adaptation, a method designed to encourage the customized models, which do not have access to each other during fine-tuning, to have orthogonal residual weights. This ensures that during inference time, the customized models can be summed with minimal interference. Our proposed method is both simple and versatile, applicable to nearly all optimizable weights in the model architecture. Through an extensive set of quantitative and qualitative evaluations, our method consistently outperforms relevant baselines in terms of efficiency and identity preservation, demonstrating a significant leap toward scalable customization of diffusion models.

Conventional image sensors digitize high-resolution images at fast frame rates, producing a large amount of data that needs to be transmitted off the sensor for fur-ther processing. This is challenging for perception system operating on edge devices, because communication is power inefficient and induces latency. Fueled by innovations in stacked image sensor fabrication, emerging sensor-processors offer programmability and processing capabilities directly on the sensor. We exploit these capabilities by developing an efficient recurrent neural network architecture, PixelRNN, that encodes spatio-temporal features on the sensor using purely binary operations. PixelRNN reduces the amount of data to be transmitted off the sensor by factors up to 256 compared to the raw sensor data while offering competitive accuracy for hand gesture recognition and lip reading tasks. We experimentally validate PixelRNN using a prototype implementation on the SCAMP-5 sensor-processor platform.

Despite recent advances in text-to-3D generative methods, there is a notable absence of reliable evaluation metrics. Existing metrics usually focus on a single criterion each, such as how well the asset aligned with the input text. These metrics lack the flexibility to generalize to different evaluation criteria and might not align well with human preferences. Conducting user preference studies is an alternative that offers both adaptability and human-aligned results. User studies, however, can be very ex-pensive to scale. This paper presents an automatic, ver-satile, and human-aligned evaluation metric for text-to-3D generative models. To this end, we first develop a prompt generator using GPT-4V to generate evaluating prompts, which serve as input to compare text-to-3D models. We further design a method instructing GPT-4V to compare two 3D assets according to user-defined crite-ria. Finally, we use these pairwise comparison results to assign these models Elo ratings. Experimental results suggest our metric strongly aligns with human preference across different evaluation criteria. Our code is available at https://github.com/3DTopia/GPTEval3D.

Thermal imaging has a variety of applications, from agricultural monitoring to building inspection to imaging under poor visibility, such as in low light, fog, and rain. However, reconstructing thermal scenes in 3D presents several challenges due to the comparatively lower resolution and limited features present in long-wave infrared (LWIR) images. To overcome these challenges, we propose a unified framework for scene reconstruction from a set of LWIR and RGB images, using a multispectral radiance field to represent a scene viewed by both visible and infrared cameras, thus leveraging information across both spectra. We calibrate the RGB and infrared cameras with respect to each other, as a preprocessing step using a simple calibration target. We demonstrate our method on real-world sets of RGB and LWIR photographs captured from a handheld thermal camera, showing the effectiveness of our method at scene representation across the visible and infrared spectra. We show that our method is capable of thermal super-resolution, as well as visually removing obstacles to reveal objects that are occluded in either the RGB or thermal channels. Please see https://yvette256.github.io/thermalnerf/ for video results as well as our code and dataset release.

We propose DMV3D, a novel 3D generation approach that uses a transformer-based 3D large reconstruction model to denoise multi-view diffusion. Our reconstruction model incorporates a triplane NeRF representation and, functioning as a denoiser, can denoise noisy multi-view images via 3D NeRF reconstruction and rendering, achieving single-stage 3D generation in the 2D diffusion denoising process. We train DMV3D on large-scale multi-view image datasets of extremely diverse objects using only image reconstruction losses, without accessing 3D assets. We demonstrate state-of-the-art results for the single-image reconstruction problem where probabilistic modeling of unseen object parts is required for generating diverse reconstructions with sharp textures. We also show high-quality text-to-3D generation results outperforming previous 3D diffusion models. Our project website is at: https://dmv3d.github.io/.

We investigate the space of weights spanned by a large collection of customized diffusion models. We populate this space by creating a dataset of over 60,000 models, each of which is a base model fine-tuned to insert a different person's visual identity. We model the underlying manifold of these weights as a subspace, which we term $\textit{weights2weights}$. We demonstrate three immediate applications of this space that result in new diffusion models -- sampling, editing, and inversion. First, sampling a set of weights from this space results in a new model encoding a novel identity. Next, we find linear directions in this space corresponding to semantic edits of the identity (e.g., adding a beard), resulting in a new model with the original identity edited. Finally, we show that inverting a single image into this space encodes a realistic identity into a model, even if the input image is out of distribution (e.g., a painting). We further find that these linear properties of the diffusion model weight space extend to other visual concepts. Our results indicate that the weight space of fine-tuned diffusion models can behave as an interpretable $\textit{meta}$-latent space producing new models.

Physically realistic materials are pivotal in augmenting the realism of 3D assets across various applications and lighting conditions. However, existing 3D assets and generative models often lack authentic material properties. Manual assignment of materials using graphic software is a tedious and time-consuming task. In this paper, we exploit advancements in Multimodal Large Language Models (MLLMs), particularly GPT-4V, to present a novel approach, Make-it-Real: 1) We demonstrate that GPT-4V can effectively recognize and describe materials, allowing the construction of a detailed material library. 2) Utilizing a combination of visual cues and hierarchical text prompts, GPT-4V precisely identifies and aligns materials with the corresponding components of 3D objects. 3) The correctly matched materials are then meticulously applied as reference for the new SVBRDF material generation according to the original albedo map, significantly enhancing their visual authenticity. Make-it-Real offers a streamlined integration into the 3D content creation workflow, showcasing its utility as an essential tool for developers of 3D assets.

Research on video generation has recently made tremendous progress, enabling high-quality videos to be generated from text prompts or images. Adding control to the video generation process is an important goal moving forward and recent approaches that condition video generation models on camera trajectories take an important step towards this goal. Yet, it remains challenging to generate a video of the same scene from multiple different camera trajectories. Solutions to this multi-video generation problem could enable large-scale 3D scene generation with editable camera trajectories, among other applications. We introduce collaborative video diffusion (CVD) as an important step towards this vision. The CVD framework includes a novel cross-video synchronization module that promotes consistency between corresponding frames of the same video rendered from different camera poses using an epipolar attention mechanism. Trained on top of a state-of-the-art camera-control module for video generation, CVD generates multiple videos rendered from different camera trajectories with significantly better consistency than baselines, as shown in extensive experiments.

Cryo-electron microscopy (cryo-EM) is an experimental technique for protein structure determination that images an ensemble of macromolecules in near-physiological contexts. While recent advances enable the reconstruction of dynamic conformations of a single biomolecular complex, current methods do not adequately model samples with mixed conformational and compositional heterogeneity. In particular, datasets containing mixtures of multiple proteins require the joint inference of structure, pose, compositional class, and conformational states for 3D reconstruction. Here, we present Hydra, an approach that models both conformational and compositional heterogeneity fully ab initio by parameterizing structures as arising from one of K neural fields. We employ a hybrid optimization strategy and demonstrate the effectiveness of our approach on synthetic datasets composed of mixtures of proteins with large degrees of conformational variability. We additionally demonstrate Hydra on an experimental dataset imaged of a cellular lysate containing a mixture of different protein complexes. Hydra expands the expressivity of heterogeneous reconstruction methods and thus broadens the scope of cryo-EM to increasingly complex samples.

Recent advances in text-to-image generation have enabled the creation of high-quality images with diverse applications. However, accurately describing desired visual attributes can be challenging, especially for non-experts in art and photography. An intuitive solution involves adopting favorable attributes from source images. Current methods attempt to distill identity and style from source images. However, "style" is a broad concept that includes texture, color, and artistic elements, but does not cover other important attributes like lighting and dynamics. Additionally, a simplified "style" adaptation prevents combining multiple attributes from different sources into one generated image. In this work, we formulate a more effective approach to decompose the aesthetics of a picture into specific visual attributes, letting users apply characteristics like lighting, texture, and dynamics from different images. To achieve this goal, we constructed the first fine-grained visual attributes dataset (FiVA) to the best of our knowledge. This FiVA dataset features a well-organized taxonomy for visual attributes and includes 1 M high-quality generated images with visual attribute annotations. Leveraging this dataset, we propose a fine-grained visual attributes adaptation framework (FiVA-Adapter) , which decouples and adapts visual attributes from one or more source images into a generated one. This approach enhances user-friendly customization, allowing users to selectively apply desired attributes to create images that meet their unique preferences and specific content requirements.

Emerging holographic display technology offers unique capabilities for next-eneration virtual reality systems. Current holographic near-eye displays, however, only support a small etendue, which results in a direct tradeoff between achievable field of view and eyebox size. Etendue expansion has recently been explored, but existing approaches are either fundamentally limited in the image quality that can be achieved or they require extremely high-speed spatial light modulators. We describe a new etendue expansion approach that combines multiple coherent sources with content-adaptive amplitude modulation of the hologram spectrum in the Fourier plane. To generate time-multiplexed phase and amplitude patterns for our spatial light modulators, we devise a pupil-aware gradient-descent-based computer-enerated holography algorithm that is supervised by a large-baseline target light field. Compared with relevant baseline approaches, ours demonstrates significant improvements in image quality and etendue in simulation and with an experimental holographic display prototype.

Symmetries are ubiquitous across all kinds of objects, whether in nature or in man-made creations. While these symmetries may seem intuitive to the human eye, detecting them with a machine is nontrivial due to the vast search space. Classical geometry-based methods work by aggregating "votes" for each symmetry but struggle with noise. In contrast, learning-based methods may be more robust to noise, but often overlook partial symmetries due to the scarcity of annotated data. In this work, we address this challenge by proposing a novel symmetry detection method that marries classical symmetry detection techniques with recent advances in generative modeling. Specifically, we apply Langevin dynamics to a redefined symmetry space to enhance robustness against noise. We provide empirical results on a variety of shapes that suggest our method is not only robust to noise, but can also identify both partial and global symmetries. Moreover, we demonstrate the utility of our detected symmetries in various downstream tasks, such as compression and symmetrization of noisy shapes.

Gaussian splatting has become a popular representation for novel-view synthesis, exhibiting clear strengths in efficiency, photometric quality, and compositional edibility. Following its success, many works have extended Gaussians to 4D, showing that dynamic Gaussians maintain these benefits while also tracking scene geometry far better than alternative representations. Yet, these methods assume dense multi-view videos as supervision. In this work, we are interested in extending the capability of Gaussian scene representations to casually captured monocular videos. We show that existing 4D Gaussian methods dramatically fail in this setup because the monocular setting is underconstrained. Building off this finding, we propose a method we call Dynamic Gaussian Marbles, which consist of three core modifications that target the difficulties of the monocular setting. First, we use isotropic Gaussian "marbles'', reducing the degrees of freedom of each Gaussian. Second, we employ a hierarchical divide and-conquer learning strategy to efficiently guide the optimization towards solutions with globally coherent motion. Finally, we add image-level and geometry-level priors into the optimization, including a tracking loss that takes advantage of recent progress in point tracking. By constraining the optimization, Dynamic Gaussian Marbles learns Gaussian trajectories that enable novel-view rendering and accurately capture the 3D motion of the scene elements. We evaluate on the Nvidia Dynamic Scenes dataset and the DyCheck iPhone dataset, and show that Gaussian Marbles significantly outperforms other Gaussian baselines in quality, and is on-par with non-Gaussian representations, all while maintaining the efficiency, compositionality, editability, and tracking benefits of Gaussians. Our project page can be found here https://geometry.stanford.edu/projects/dynamic-gaussian-marbles.github.io/.

The 2024 IEEE VGTC Virtual Reality Technical Achievement Award goes to Gordon Wetzstein of the Department of Electrical Engineering at Stanford University, in recognition of his pioneering research in the development of computational display and imaging technologies.

Capturing images is a key part of automation for high-level tasks such as scene text recognition. Low-light conditions pose a challenge for high-level perception stacks, which are often optimized on well-lit, artifact-free images. Reconstruction methods for low-light images can produce well-lit counterparts, but typically at the cost of high-frequency details critical for downstream tasks. We propose Diffusion in the Dark (DiD), a diffusion model for low-light image reconstruction for text recognition. DiD provides qualitatively competitive reconstructions with that of state-of-the-art (SOTA), while preserving high-frequency details even in extremely noisy, dark conditions. We demonstrate that DiD, without any task-specific optimization, can outperform SOTA low-light methods in low-light text recognition on real images, bolstering the potential of diffusion models to solve ill-posed inverse problems. Our code and pretrained models can be found on https://ccnguyen.github.io/diffusion-in-the-dark/.

Abstract The field of visual computing is rapidly advancing due to the emergence of generative artificial intelligence (AI), which unlocks unprecedented capabilities for the generation, editing, and reconstruction of images, videos, and 3D scenes. In these domains, diffusion models are the generative AI architecture of choice. Within the last year alone, the literature on diffusion‐based tools and applications has seen exponential growth and relevant papers are published across the computer graphics, computer vision, and AI communities with new works appearing daily on arXiv. This rapid growth of the field makes it difficult to keep up with all recent developments. The goal of this state‐of‐the‐art report (STAR) is to introduce the basic mathematical concepts of diffusion models, implementation details and design choices of the popular Stable Diffusion model, as well as overview important aspects of these generative AI tools, including personalization, conditioning, inversion, among others. Moreover, we give a comprehensive overview of the rapidly growing literature on diffusion‐based generation and editing, categorized by the type of generated medium, including 2D images, videos, 3D objects, locomotion, and 4D scenes. Finally, we discuss available datasets, metrics, open challenges, and social implications. This STAR provides an intuitive starting point to explore this exciting topic for researchers, artists, and practitioners alike.

Abstract Creating plausible motions for a diverse range of characters is a long‐standing goal in computer graphics. Current learning‐based motion synthesis methods rely on large‐scale motion datasets, which are often difficult if not impossible to acquire. On the other hand, pose data is more accessible, since static posed characters are easier to create and can even be extracted from images using recent advancements in computer vision. In this paper, we tap into this alternative data source and introduce a neural motion synthesis approach through retargeting, which generates plausible motion of various characters that only have pose data by transferring motion from one single existing motion capture dataset of another drastically different characters. Our experiments show that our method effectively combines the motion features of the source character with the pose features of the target character, and performs robustly with small or noisy pose data sets, ranging from a few artist‐created poses to noisy poses estimated directly from images. Additionally, a conducted user study indicated that a majority of participants found our retargeted motion to be more enjoyable to watch, more lifelike in appearance, and exhibiting fewer artifacts. Our code and dataset can be accessed here.

Diffusion models offer unprecedented image generation power given just a text prompt. While emerging approaches for controlling diffusion models have enabled users to specify the desired spatial layouts of the generated content, they cannot predict or control where viewers will pay more attention due to the complexity of human vision. Recognizing the significance of attention-controllable image generation in practical applications, we present a saliency-guided framework to incorporate the data priors of human visual attention mechanisms into the generation process. Given a user-specified viewer attention distribution, our control module conditions a diffusion model to generate images that attract viewers’ attention toward the desired regions. To assess the efficacy of our approach, we performed an eye-tracked user study and a large-scale model-based saliency analysis. The results evidence that both the cross-user eye gaze distributions and the saliency models’ predictions align with the desired attention distributions. Lastly, we outline several applications, including interactive design of saliency guidance, attention suppression in unwanted regions, and adaptive generation for varied display/viewing conditions.

Holographic near-eye displays are a promising technology to solve long-standing challenges in virtual and augmented reality display systems. Over the last few years, many different computer-generated holography (CGH) algorithms have been proposed that are supervised by different types of target content, such as 2.5D RGB-depth maps, 3D focal stacks, and 4D light fields. It is unclear, however, what the perceptual implications are of the choice of algorithm and target content type. In this work, we build a perceptual testbed of a full-color, high-quality holographic near-eye display. Under natural viewing conditions, we examine the effects of various CGH supervision formats and conduct user studies to assess their perceptual impacts on 3D realism. Our results indicate that CGH algorithms designed for specific viewpoints exhibit noticeable deficiencies in achieving 3D realism. In contrast, holograms incorporating parallax cues consistently outperform other formats across different viewing conditions, including the center of the eyebox. This finding is particularly interesting and suggests that the inclusion of parallax cues in CGH rendering plays a crucial role in enhancing the overall quality of the holographic experience. This work represents an initial stride towards delivering a perceptually realistic 3D experience with holographic near-eye displays.

Generative models have shown great promise in synthesizing photorealistic 3D objects, but they require large amounts of training data. We introduce SinGRAF, a 3D-aware generative model that is trained with a few input images of a single scene. Once trained, SinGRAF generates different realizations of this 3D scene that preserve the appearance of the input while varying scene layout. For this purpose, we build on recent progress in 3D GAN architectures and introduce a novel progressive-scale patch discrimination approach during training. With several experiments, we demonstrate that the results produced by Sin-GRAF outperform the closest related works in both quality and diversity by a large margin.

Recent advances in neural radiance fields have enabled the high-fidelity 3D reconstruction of complex scenes for novel view synthesis. However, it remains underexplored how the appearance of such representations can be efficiently edited while maintaining photorealism. In this work, we present PaletteNeRF, a novel method for photorealistic appearance editing of neural radiance fields (NeRF) based on 3D color decomposition. Our method decom-poses the appearance of each 3D point into a linear combination of palette-based bases (i.e., 3D segmentations defined by a group of NeRF-type functions) that are shared across the scene. While our palette-based bases are view-independent, we also predict a view-dependent function to capture the color residual (e.g., specular shading). During training, we jointly optimize the basis functions and the color palettes, and we also introduce novel regulariz-ers to encourage the spatial coherence of the decomposition. Our method allows users to efficiently edit the appearance of the 3D scene by modifying the color palettes. We also extend our framework with compressed semantic features for semantic-aware appearance editing. We demonstrate that our technique is superior to baseline methods both quantitatively and qualitatively for appearance editing of complex real-world scenes. Our project page is https://palettenerf.github.io.

Diffusion models have emerged as the state-of-the-art for image generation, among other tasks. Here, we present an efficient diffusion-based model for 3D-aware generation of neural fields. Our approach pre-processes training data, such as ShapeNet meshes, by converting them to continuous occupancy fields and factoring them into a set of axis-aligned triplane feature representations. Thus, our 3D training scenes are all represented by 2D feature planes, and we can directly train existing 2D diffusion models on these representations to generate 3D neural fields with high quality and diversity, outperforming alternative approaches to 3D-aware generation. Our approach requires essential modifications to existing triplane factorization pipelines to make the resulting features easy to learn for the diffusion model. We demonstrate state-of-the-art results on 3D generation on several object classes from ShapeNet.

This work introduces alternating latent topologies (ALTO) for high-fidelity reconstruction of implicit 3D surfaces from noisy point clouds. Previous work identifies that the spatial arrangement of latent encodings is important to recover detail. One school of thought is to encode a latent vector for each point (point latents). Another school of thought is to project point latents into a grid (grid latents) which could be a voxel grid or triplane grid. Each school of thought has tradeoffs. Grid latents are coarse and lose high-frequency detail. In contrast, point latents preserve detail. However, point latents are more difficult to decode into a surface, and quality and runtime suffer. In this paper, we propose ALTO to sequentially alternate between geometric representations, before converging to an easy-to-decode latent. We find that this preserves spatial expressiveness and makes decoding lightweight. We validate ALTO on implicit 3D recovery and observe not only a performance improvement over the state-of-the-art, but a runtime improvement of$3-10\times$. Project website at https://visual.ee.ucla.edu/alto.htm/.

The ability to create realistic animatable and relightable head avatars from casual video sequences would open up wide ranging applications in communication and entertainment. Current methods either build on explicit 3D morphable meshes (3DMM) or exploit neural implicit representations. The former are limited by fixed topology, while the latter are non-trivial to deform and inefficient to render. Furthermore, existing approaches entangle lighting and albedo, limiting the ability to re-render the avatar in new environments. In contrast, we propose PointAvatar, a deformable point-based representation that disentangles the source color into intrinsic albedo and normal-dependent shading. We demonstrate that PointAvatar bridges the gap between existing mesh- and implicit representations, combining high-quality geometry and appearance with topological flexibility, ease of deformation and rendering efficiency. We show that our method is able to generate animatable 3D avatars using monocular videos from multiple sources including hand-held smartphones, laptop webcams and internet videos, achieving state-of-the-art quality in challenging cases where previous methods fail, e.g., thin hair strands, while being significantly more efficient in training than competing methods.

In this work, we introduce CC3D, a conditional generative model that synthesizes complex 3D scenes conditioned on 2D semantic scene layouts, trained using single-view images. Different from most existing 3D GANs that limit their applicability to aligned single objects, we focus on generating complex scenes with multiple objects, by modeling the compositional nature of 3D scenes. By devising a 2D layout-based approach for 3D synthesis and implementing a new 3D field representation with a stronger geometric inductive bias, we have created a 3D GAN that is both efficient and of high quality, while allowing for a more controllable generation process. Our evaluations on synthetic 3D-FRONT and real-world KITTI-360 datasets demonstrate that our model generates scenes of improved visual and geometric quality in comparison to previous works.

Scene extrapolation—the idea of generating novel views by flying into a given image—is a promising, yet challenging task. For each predicted frame, a joint inpainting and 3D refinement problem has to be solved, which is ill posed and includes a high level of ambiguity. Moreover, training data for long-range scenes is difficult to obtain and usually lacks sufficient views to infer accurate camera poses. We introduce DiffDreamer, an unsupervised framework capable of synthesizing novel views depicting a long camera trajectory while training solely on internet-collected images of nature scenes. Utilizing the stochastic nature of the guided denoising steps, we train the diffusion models to refine projected RGBD images but condition the denoising steps on multiple past and future frames for inference. We demonstrate that image-conditioned diffusion models can effectively perform long-range scene extrapolation while preserving consistency significantly better than prior GAN-based methods. DiffDreamer is a powerful and efficient solution for scene extrapolation, producing impressive results despite limited supervision. Project page: https://primecai.github.io/diffdreamer.

We present a diffusion-based model for 3D-aware generative novel view synthesis from as few as a single input image. Our model samples from the distribution of possible renderings consistent with the input and, even in the presence of ambiguity, is capable of rendering diverse and plausible novel views. To achieve this, our method makes use of existing 2D diffusion backbones but, crucially, incorporates geometry priors in the form of a 3D feature volume. This latent feature field captures the distribution over possible scene representations and improves our method’s ability to generate view-consistent novel renderings. In addition to generating novel views, our method has the ability to autoregressively synthesize 3D-consistent sequences. We demonstrate state-of-the-art results on synthetic renderings and room-scale scenes; we also show compelling results for challenging, real-world objects.

We introduce PointOdyssey, a large-scale synthetic dataset, and data generation framework, for the training and evaluation of long-term fine-grained tracking algorithms. Our goal is to advance the state-of-the-art by placing emphasis on long videos with naturalistic motion. Toward the goal of naturalism, we animate deformable characters using real-world motion capture data, we build 3D scenes to match the motion capture environments, and we render camera viewpoints using trajectories mined via structure-from-motion on real videos. We create combinatorial diversity by randomizing character appearance, motion profiles, materials, lighting, 3D assets, and atmospheric effects. Our dataset currently includes 104 videos, averaging 2,000 frames long, with orders of magnitude more correspondence annotations than prior work. We show that existing methods can be trained from scratch in our dataset and outperform the published variants. Finally, we introduce modifications to the PIPs point tracking method, greatly widening its temporal receptive field, which improves its performance on PointOdyssey as well as on two real-world benchmarks. Our data and code are publicly available at: https://pointodyssey.com

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.

This work introduces off-axis layered displays, the first approach to stereoscopic direct-view displays with support for focus cues. Off-axis layered displays combine a head-mounted display with a traditional direct-view display for encoding a focal stack and thus, for providing focus cues. To explore the novel display architecture, we present a complete processing pipeline for the real-time computation and post-render warping of off-axis display patterns. In addition, we build two prototypes using a head-mounted display in combination with a stereoscopic direct-view display, and a more widely available monoscopic direct-view display. In addition we show how extending off-axis layered displays with an attenuation layer and with eye-tracking can improve image quality. We thoroughly analyze each component in a technical evaluation and present examples captured through our prototypes.

Unsupervised generation of high-quality multi-view-consistent images and 3D shapes using only collections of single-view 2D photographs has been a long-standing challenge. Existing 3D GANs are either compute intensive or make approximations that are not 3D-consistent; the former limits quality and resolution of the generated images and the latter adversely affects multi-view consistency and shape quality. In this work, we improve the computational efficiency and image quality of 3D GANs without overly relying on these approximations. We introduce an expressive hybrid explicit implicit network architecture that, together with other design choices, synthesizes not only high-resolution multi-view-consistent images in real time but also produces high-quality 3D geometry. By decoupling feature generation and neural rendering, our framework is able to leverage state-of-the-art 2D CNN generators, such as StyleGAN2, and inherit their efficiency and expressiveness. We demonstrate state-of-the-art 3D-aware synthesis with FFHQ and AFHQ Cats, among other experiments.

Coordinate-based networks have emerged as a powerful tool for 3D representation and scene reconstruction. These networks are trained to map continuous input coordinates to the value of a signal at each point. Still, current architectures are black boxes: their spectral characteristics cannot be easily analyzed, and their behavior at unsupervised points is difficult to predict. Moreover, these networks are typically trained to represent a signal at a single scale, so naive downsampling or upsampling results in artifacts. We introduce band-limited coordinate networks (BACON), a network architecture with an analytical Fourier spectrum. Bacon has constrained behavior at unsupervised points, can be designed based on the spectral characteristics of the represented signal, and can represent signals at multiple scales without per-scale supervision. We demonstrate Bacon for multiscale neural representation of images, radiance fields, and 3D scenes using signed distance functions and show that it outperforms conventional single-scale coordinate networks in terms of interpretability and quality.

Computationally removing the motion blur introduced by camera shake or object motion in a captured image remains a challenging task in computational photography. Deblurring methods are often limited by the fixed global exposure time of the image capture process. The post-processing algorithm either must deblur a longer exposure that contains relatively little noise or denoise a short exposure that intentionally removes the opportunity for blur at the cost of increased noise. We present a novel approach of leveraging spatially varying pixel exposures for motion deblurring using next-generation focal-plane sensor-processors along with an end-to-end design of these exposures and a machine learning-based motion-deblurring framework. We demonstrate in simulation and a physical prototype that learned spatially varying pixel exposures (L-SVPE) can successfully deblur scenes while recovering high frequency detail. Our work illustrates the promising role that focal-plane sensor-processors can play in the future of computational imaging.

The ability to image high-dynamic-range (HDR) scenes is crucial in many computer vision applications. The dynamic range of conventional sensors, however, is fundamentally limited by their well capacity, resulting in saturation of bright scene parts. To overcome this limitation, emerging sensors offer in-pixel processing capabilities to encode the incident irradiance. Among the most promising encoding schemes is modulo wrapping, which results in a computational photography problem where the HDR scene is computed by an irradiance unwrapping algorithm from the wrapped low-dynamic-range (LDR) sensor image. Here, we design a neural network-based algorithm that outperforms previous irradiance unwrapping methods and we design a perceptually inspired “mantissa,” or log-modulo, encoding scheme that more efficiently wraps an HDR scene into an LDR sensor. Combined with our reconstruction framework, MantissaCam achieves state-of-the-art results among modulo-type snapshot HDR imaging approaches. We demonstrate the efficacy of our method in simulation and show benefits of our algorithm on modulo images captured with a prototype implemented with a programmable sensor.

Learned graph neural networks (GNNs) have recently been established as fast and accurate alternatives for principled solvers in simulating the dynamics of physical systems. In many application domains across science and engineering, however, we are not only interested in a forward simulation but also in solving inverse problems with constraints defined by a partial differential equation (PDE). Here we explore GNNs to solve such PDE-constrained inverse problems. Given a sparse set of measurements, we are interested in recovering the initial condition or parameters of the PDE. We demonstrate that GNNs combined with autodecoder-style priors are well-suited for these tasks, achieving more accurate estimates of initial conditions or physical parameters than other learned approaches when applied to the wave equation or Navier Stokes equations. We also demonstrate computational speedups of up to 90x using GNNs compared to principled solvers.

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.

Cryo-electron microscopy (cryo-EM) is an imaging modality that provides unique insights into the dynamics of proteins and other building blocks of life. The algorithmic challenge of jointly estimating the poses, 3D structure, and conformational heterogeneity of a biomolecule from millions of noisy and randomly oriented 2D projections in a computationally efficient manner, however, remains unsolved. Our method, cryoFIRE, performs ab initio heterogeneous reconstruction with unknown poses in an amortized framework, thereby avoiding the computationally expensive step of pose search while enabling the analysis of conformational heterogeneity. Poses and conformation are jointly estimated by an encoder while a physics-based decoder aggregates the images into an implicit neural representation of the conformational space. We show that our method can provide one order of magnitude speedup on datasets containing millions of images, without any loss of accuracy. We validate that the joint estimation of poses and conformations can be amortized over the size of the dataset. For the first time, we prove that an amortized method can extract interpretable dynamic information from experimental datasets.

Neural representations are popular for representing shapes as they can be used for data cleanup, model completion, shape editing, and shape synthesis. Current neural representations can be categorized as either overfitting to a single object instance, or representing a collection of objects. However, neither allows accurate editing of neural scene representations: on the one hand, methods that overfit objects achieve highly accurate reconstructions but do not support editing, as they do not generalize to unseen object configurations; on the other hand, methods that represent a family of objects with variations do generalize but produce approximate reconstructions. We propose NeuForm to combine the advantages of both overfitted and generalizable representations by adaptively overfitting a generalizable representation to regions where reliable data is available, while using the generalizable representation everywhere else. We achieve this with a carefully designed architecture and an approach that blends the network weights of the two representations. We demonstrate edits that successfully reconfigure parts of human-made shapes, such as chairs, tables, and lamps, while preserving the accuracy of an overfitted shape representation. We compare with two state-of-the-art competitors and demonstrate clear improvements in terms of plausibility and fidelity of the resultant edits.

Abstract Synthesizing photo‐realistic images and videos is at the heart of computer graphics and has been the focus of decades of research. Traditionally, synthetic images of a scene are generated using rendering algorithms such as rasterization or ray tracing, which take specifically defined representations of geometry and material properties as input. Collectively, these inputs define the actual scene and what is rendered, and are referred to as the scene representation (where a scene consists of one or more objects). Example scene representations are triangle meshes with accompanied textures (e.g., created by an artist), point clouds (e.g., from a depth sensor), volumetric grids (e.g., from a CT scan), or implicit surface functions (e.g., truncated signed distance fields). The reconstruction of such a scene representation from observations using differentiable rendering losses is known as inverse graphics or inverse rendering. Neural rendering is closely related, and combines ideas from classical computer graphics and machine learning to create algorithms for synthesizing images from real‐world observations. Neural rendering is a leap forward towards the goal of synthesizing photo‐realistic image and video content. In recent years, we have seen immense progress in this field through hundreds of publications that show different ways to inject learnable components into the rendering pipeline. This state‐of‐the‐art report on advances in neural rendering focuses on methods that combine classical rendering principles with learned 3D scene representations, often now referred to as neural scene representations. A key advantage of these methods is that they are 3D‐consistent by design, enabling applications such as novel viewpoint synthesis of a captured scene. In addition to methods that handle static scenes, we cover neural scene representations for modeling non‐rigidly deforming objects and scene editing and composition. While most of these approaches are scene‐specific, we also discuss techniques that generalize across object classes and can be used for generative tasks. In addition to reviewing these state‐of‐the‐art methods, we provide an overview of fundamental concepts and definitions used in the current literature. We conclude with a discussion on open challenges and social implications.

This work introduces the first approach to video see-through mixed reality with full support for focus cues. By combining the flexibility to adjust the focus distance found in varifocal designs with the robustness to eye-tracking error found in multifocal designs, our novel display architecture reliably delivers focus cues over a large workspace. In particular, we introduce gaze-contingent layered displays and mixed reality focal stacks, an efficient representation of mixed reality content that lends itself to fast processing for driving layered displays in real time. We thoroughly evaluate this approach by building a complete end-to-end pipeline for capture, render, and display of focus cues in video see-through displays that uses only off-the-shelf hardware and compute components.

Understanding and modeling the dynamics of human gaze behavior in 360° environments is crucial for creating, improving, and developing emerging virtual reality applications. However, recruiting human observers and acquiring enough data to analyze their behavior when exploring virtual environments requires complex hardware and software setups, and can be time-consuming. Being able to generate virtual observers can help overcome this limitation, and thus stands as an open problem in this medium. Particularly, generative adversarial approaches could alleviate this challenge by generating a large number of scanpaths that reproduce human behavior when observing new scenes, essentially mimicking virtual observers. However, existing methods for scanpath generation do not adequately predict realistic scanpaths for 360° images. We present ScanGAN360, a new generative adversarial approach to address this problem. We propose a novel loss function based on dynamic time warping and tailor our network to the specifics of 360° images. The quality of our generated scanpaths outperforms competing approaches by a large margin, and is almost on par with the human baseline. ScanGAN360 allows fast simulation of large numbers of virtual observers, whose behavior mimics real users, enabling a better understanding of gaze behavior, facilitating experimentation, and aiding novel applications in virtual reality and beyond.

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.

Numerical integration is a foundational technique in scientific computing and is at the core of many computer vision applications. Among these applications, neural volume rendering has recently been proposed as a new paradigm for view synthesis, achieving photorealistic image quality. However, a fundamental obstacle to making these methods practical is the extreme computational and memory requirements caused by the required volume integrations along the rendered rays during training and inference. Millions of rays, each requiring hundreds of forward passes through a neural network are needed to approximate those integrations with Monte Carlo sampling. Here, we propose automatic integration, a new framework for learning efficient, closed-form solutions to integrals using coordinate-based neural networks. For training, we instantiate the computational graph corresponding to the derivative of the coordinate-based network. The graph is fitted to the signal to integrate. After optimization, we reassemble the graph to obtain a network that represents the antiderivative. By the fundamental theorem of calculus, this enables the calculation of any definite integral in two evaluations of the network. Applying this approach to neural rendering, we improve a tradeoff between rendering speed and image quality: improving render times by greater than 10× with a tradeoff of reduced image quality.

Convolutional neural networks (CNN) have emerged as a powerful tool for solving computational imaging reconstruction problems. However, CNNs are generally difficult-to-understand black-boxes. Accordingly, it is challenging to know when they will work and, more importantly, when they will fail. This limitation is a major barrier to their use in safety-critical applications like medical imaging: Is that blob in the reconstruction an artifact or a tumor?In this work we use Stein’s unbiased risk estimate (SURE) to develop per-pixel confidence intervals, in the form of heatmaps, for compressive sensing reconstruction using the approximate message passing (AMP) framework with CNN-based denoisers. These heatmaps tell end-users how much to trust an image formed by a CNN, which could greatly improve the utility of CNNs in various computational imaging applications.

This paper introduces and solves the simultaneous source separation and phase retrieval (S3PR) problem. S3PR is an important but largely unsolved problem in a number application domains, including microscopy, wireless communication, and imaging through scattering media, where one has multiple independent coherent sources whose phase is difficult to measure. In general, S3PR is highly under-determined, non-convex, and difficult to solve. In this work, we demonstrate that by restricting the solutions to lie in the range of a deep generative model, we can constrain the search space sufficiently to solve S3PR.Code associated with this work is available at https://github.com/computational-imaging/DeepS3PR. An extended version of this work is available at https://arxiv.org/abs/2002.05856.

Plug and play (P&P) algorithms iteratively apply highly optimized image denoisers to impose priors and solve computational image reconstruction problems, to great effect. However, in general the "effective noise", that is the difference between the true signal and the intermediate solution, within the iterations of P&P algorithms is neither Gaussian nor white. This fact makes existing denoising algorithms suboptimal.In this work, we propose a CNN architecture for removing colored Gaussian noise and combine it with the recently proposed VDAMP algorithm, whose effective noise follows a predictable colored Gaussian distribution. We apply the resulting denoising-based VDAMP (D-VDAMP) algorithm to variable density sampled compressive MRI where it substantially outperforms existing techniques.

Monocular depth estimation remains a challenging problem, despite significant advances in neural network architectures that leverage pictorial depth cues alone. Inspired by depth from defocus and emerging point spread function engineering approaches that optimize programmable optics end-to-end with depth estimation networks, we propose a new and improved framework for depth estimation from a single RGB image using a learned phase-coded aperture. Our optimized aperture design uses rotational symmetry constraints for computational efficiency, and we jointly train the optics and the network using an occlusion-aware image formation model that provides more accurate defocus blur at depth discontinuities than previous techniques do. Using this framework and a custom prototype camera, we demonstrate state-of-the art image and depth estimation quality among end-to-end optimized computational cameras in simulation and experiment.

Imaging depth and spectrum have been extensively studied in isolation from each other for decades. Recently, hyperspectral-depth (HS-D) imaging emerges to capture both information simultaneously by combining two different imaging systems; one for depth, the other for spectrum. While being accurate, this combinational approach induces increased form factor, cost, capture time, and alignment/registration problems. In this work, departing from the combinational principle, we propose a compact single-shot monocular HS-D imaging method. Our method uses a diffractive optical element (DOE), the point spread function of which changes with respect to both depth and spectrum. This enables us to reconstruct spectrum and depth from a single captured image. To this end, we develop a differentiable simulator and a neural-network-based reconstruction method that are jointly optimized via automatic differentiation. To facilitate learning the DOE, we present a first HS-D dataset by building a benchtop HS-D imager that acquires high-quality ground truth. We evaluate our method with synthetic and real experiments by building an experimental prototype and achieve state-of-the-art HS-D imaging results.

Compressive imaging using coded apertures (CA) is a powerful technique that can be used to recover depth, light fields, hyperspectral images and other quantities from a single snapshot. The performance of compressive imaging systems based on CAs mostly depends on two factors: the properties of the mask's attenuation pattern, that we refer to as "codification", and the computational techniques used to recover the quantity of interest from the coded snapshot. In this work, we introduce the idea of using time-varying CAs synchronized with spatially varying pixel shutters. We divide the exposure of a sensor into sub-exposures at the beginning of which the CA mask changes and at which the sensor's pixels are simultaneously and individually switched "on" or "off". This is a practically appealing codification as it does not introduce additional optical components other than the already present CA but uses a change in the pixel shutter that can be easily realized electronically. We show that our proposed time-multiplexed coded aperture (TMCA) can be optimized end to end and induces better coded snapshots enabling superior reconstructions in two different applications: compressive light field imaging and hyperspectral imaging. We demonstrate both in simulation and with real captures (taken with prototypes we built) that this codification outperforms the state-of-the-art compressive imaging systems by a large margin in those applications.

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.

Holographic near-eye displays promise unprecedented capabilities for virtual and augmented reality (VR/AR) systems. The image quality achieved by current holographic displays, however, is limited by the wave propagation models used to simulate the physical optics. We propose a neural network-parameterized plane-to-multiplane wave propagation model that closes the gap between physics and simulation. Our model is automatically trained using camera feedback and it outperforms related techniques in 2D plane-to-plane settings by a large margin. Moreover, it is the first network-parameterized model to naturally extend to 3D settings, enabling high-quality 3D computer-generated holography using a novel phase regularization strategy of the complex-valued wave field. The efficacy of our approach is demonstrated through extensive experimental evaluation with both VR and optical see-through AR display prototypes.

Real-world, imaging systems acquire measurements that are degraded by noise, optical aberrations, and other imperfections that make image processing for human viewing and higher-level perception tasks challenging. Conventional cameras address this problem by compartmentalizing imaging from high-level task processing. As such, conventional imaging involves processing the RAW sensor measurements in a sequential pipeline of steps, such as demosaicking, denoising, deblurring, tone-mapping, and compression. This pipeline is optimized to obtain a visually pleasing image. High-level processing, however, involves steps such as feature extraction, classification, tracking, and fusion. While this silo-ed design approach allows for efficient development, it also dictates compartmentalized performance metrics without knowledge of the higher-level task of the camera system. For example, today’s demosaicking and denoising algorithms are designed using perceptual image quality metrics but not with domain-specific tasks such as object detection in mind. We propose an end-to-end differentiable architecture that jointly performs demosaicking, denoising, deblurring, tone-mapping, and classification (see Figure 1). The architecture does not require any intermediate losses based on perceived image quality and learns processing pipelines whose outputs differ from those of existing ISPs optimized for perceptual quality, preserving fine detail at the cost of increased noise and artifacts. We show that state-of-the-art ISPs discard information that is essential in corner cases, such as extremely low-light conditions, where conventional imaging and perception stacks fail. We demonstrate on captured and simulated data that our model substantially improves perception in low light and other challenging conditions, which is imperative for real-world applications such as autonomous driving, robotics, and surveillance. Finally, we found that the proposed model also achieves state-of-the-art accuracy when optimized for image reconstruction in low-light conditions, validating the architecture itself as a potentially useful drop-in network for reconstruction and analysis tasks beyond the applications demonstrated in this work. Our proposed models, datasets, and calibration data are available at https://github.com/princeton-computational-imaging/DirtyPixels .

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.

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.

The cameras in modern gaze-tracking systems suffer from fundamental bandwidth and power limitations, constraining data acquisition speed to 300 Hz realistically. This obstructs the use of mobile eye trackers to perform, e.g., low latency predictive rendering, or to study quick and subtle eye motions like microsaccades using head-mounted devices in the wild. Here, we propose a hybrid frame-event-based near-eye gaze tracking system offering update rates beyond 10,000 Hz with an accuracy that matches that of high-end desktop-mounted commercial trackers when evaluated in the same conditions. Our system, previewed in Figure 1, builds on emerging event cameras that simultaneously acquire regularly sampled frames and adaptively sampled events. We develop an online 2D pupil fitting method that updates a parametric model every one or few events. Moreover, we propose a polynomial regressor for estimating the point of gaze from the parametric pupil model in real time. Using the first event-based gaze dataset, we demonstrate that our system achieves accuracies of 0.45°-1.75° for fields of view from 45° to 98°. With this technology, we hope to enable a new generation of ultra-low-latency gaze-contingent rendering and display techniques for virtual and augmented reality.

The recent success of implicit neural scene representations has presented a viable new method for how we capture and store 3D scenes. Unlike conventional 3D representations, such as point clouds, which explicitly store scene properties in discrete, localized units, these implicit representations encode a scene in the weights of a neural network which can be queried at any coordinate to produce these same scene properties. Thus far, implicit representations have primarily been optimized to estimate only the appearance and/or 3D geometry information in a scene. We take the next step and demonstrate that an existing implicit representation (SRNs) [67] is actually multi-modal; it can be further leveraged to perform per-point semantic segmentation while retaining its ability to represent appearance and geometry. To achieve this multi-modal behavior, we utilize a semi-supervised learning strategy atop the existing pre-trained scene representation. Our method is simple, general, and only requires a few tens of labeled 2D segmentation masks in order to achieve dense 3D semantic segmentation. We explore two novel applications for this semantically aware implicit neural scene representation: 3D novel view and semantic label synthesis given only a single input RGB image or 2D label mask, as well as 3D interpolation of appearance and semantics.

High-dynamic-range (HDR) imaging is crucial for many applications. Yet, acquiring HDR images with a single shot remains a challenging problem. Whereas modern deep learning approaches are successful at hallucinating plausible HDR content from a single low-dynamic-range (LDR) image, saturated scene details often cannot be faithfully recovered. Inspired by recent deep optical imaging approaches, we interpret this problem as jointly training an optical encoder and electronic decoder where the encoder is parameterized by the point spread function (PSF) of the lens, the bottleneck is the sensor with a limited dynamic range, and the decoder is a convolutional neural network (CNN). The lens surface is then jointly optimized with the CNN in a training phase; we fabricate this optimized optical element and attach it as a hardware add-on to a conventional camera during inference. In extensive simulations and with a physical prototype, we demonstrate that this end-to-end deep optical imaging approach to single-shot HDR imaging outperforms both purely CNN-based approaches and other PSF engineering approaches.

We propose a joint albedo-normal approach to non-line-of-sight (NLOS) surface reconstruction using the directional light-cone transform (D-LCT). While current NLOS imaging methods reconstruct either the albedo or surface normals of the hidden scene, the two quantities provide complementary information of the scene, so an efficient method to estimate both simultaneously is desirable. We formulate the recovery of the two quantities as a vector deconvolution problem, and solve it via Cholesky-Wiener decomposition. We demonstrate that surfaces fitted non-parametrically using our recovered normals are more accurate than those produced with NLOS surface reconstruction methods recently proposed, and are 1,000 times faster to compute than using inverse rendering.

Current LiDAR systems are limited in their ability to capture dense 3D point clouds. To overcome this challenge, deep learning-based depth completion algorithms have been developed to inpaint missing depth guided by an RGB image. However, these methods fail for low sampling rates. Here, we propose an adaptive sampling scheme for LiDAR systems that demonstrates state-of-the-art performance for depth completion at low sampling rates. Our system is fully differentiable, allowing the sparse depth sampling and the depth inpainting components to be trained end-to-end with an upstream task.

Neural implicit shape representations are an emerging paradigm that offers many potential benefits over conventional discrete representations, including memory efficiency at a high spatial resolution. Generalizing across shapes with such neural implicit representations amounts to learning priors over the respective function space and enables geometry reconstruction from partial or noisy observations. Existing generalization methods rely on conditioning a neural network on a low-dimensional latent code that is either regressed by an encoder or jointly optimized in the auto-decoder framework. Here, we formalize learning of a shape space as a meta-learning problem and leverage gradient-based meta-learning algorithms to solve this task. We demonstrate that this approach performs on par with auto-decoder based approaches while being an order of magnitude faster at test-time inference. We further demonstrate that the proposed gradient-based method outperforms encoder-decoder based methods that leverage pooling-based set encoders.

Implicitly defined, continuous, differentiable signal representations parameterized by neural networks have emerged as a powerful paradigm, offering many possible benefits over conventional representations. However, current network architectures for such implicit neural representations are incapable of modeling signals with fine detail, and fail to represent a signal's spatial and temporal derivatives, despite the fact that these are essential to many physical signals defined implicitly as the solution to partial differential equations. We propose to leverage periodic activation functions for implicit neural representations and demonstrate that these networks, dubbed sinusoidal representation networks or SIRENs, are ideally suited for representing complex natural signals and their derivatives. We analyze SIREN activation statistics to propose a principled initialization scheme and demonstrate the representation of images, wavefields, video, sound, and their derivatives. Further, we show how SIRENs can be leveraged to solve challenging boundary value problems, such as particular Eikonal equations (yielding signed distance functions), the Poisson equation, and the Helmholtz and wave equations. Lastly, we combine SIRENs with hypernetworks to learn priors over the space of SIREN functions.

Abstract Efficient rendering of photo‐realistic virtual worlds is a long standing effort of computer graphics. Modern graphics techniques have succeeded in synthesizing photo‐realistic images from hand‐crafted scene representations. However, the automatic generation of shape, materials, lighting, and other aspects of scenes remains a challenging problem that, if solved, would make photo‐realistic computer graphics more widely accessible. Concurrently, progress in computer vision and machine learning have given rise to a new approach to image synthesis and editing, namely deep generative models. Neural rendering is a new and rapidly emerging field that combines generative machine learning techniques with physical knowledge from computer graphics, e.g., by the integration of differentiable rendering into network training. With a plethora of applications in computer graphics and vision, neural rendering is poised to become a new area in the graphics community, yet no survey of this emerging field exists. This state‐of‐the‐art report summarizes the recent trends and applications of neural rendering. We focus on approaches that combine classic computer graphics techniques with deep generative models to obtain controllable and photorealistic outputs. Starting with an overview of the underlying computer graphics and machine learning concepts, we discuss critical aspects of neural rendering approaches. Specifically, our emphasis is on the type of control, i.e., how the control is provided, which parts of the pipeline are learned, explicit vs. implicit control, generalization, and stochastic vs. deterministic synthesis. The second half of this state‐of‐the‐art report is focused on the many important use cases for the described algorithms such as novel view synthesis, semantic photo manipulation, facial and body reenactment, relighting, free‐viewpoint video, and the creation of photo‐realistic avatars for virtual and augmented reality telepresence. Finally, we conclude with a discussion of the social implications of such technology and investigate open research problems.

Camera sensors rely on global or rolling shutter functions to expose an image. This fixed function approach severely limits the sensors' ability to capture high-dynamic-range (HDR) scenes and resolve high-speed dynamics. Spatially varying pixel exposures have been introduced as a powerful computational photography approach to optically encode irradiance on a sensor and computationally recover additional information of a scene, but existing approaches rely on heuristic coding schemes and bulky spatial light modulators to optically implement these exposure functions. Here, we introduce neural sensors as a methodology to optimize per-pixel shutter functions jointly with a differentiable image processing method, such as a neural network, in an end-to-end fashion. Moreover, we demonstrate how to leverage emerging programmable and re-configurable sensor-processors to implement the optimized exposure functions directly on the sensor. Our system takes specific limitations of the sensor into account to optimize physically feasible optical codes and we evaluate its performance for snapshot HDR and high-speed compressive imaging both in simulation and experimentally with real scenes.

Modern 3D printers are capable of printing large-size light-field displays at high-resolutions. However, optimizing such displays in full 3D volume for a given light-field imagery is still a challenging task. Existing light field displays optimize over relatively small resolutions using a few co-planar layers in a 2.5D fashion to keep the problem tractable. In this paper, we propose a novel end-to-end optimization approach that encodes input light field imagery as a continuous-space implicit representation in a neural network. This allows fabricating high-resolution, attenuation-based volumetric displays that exhibit the target light fields. In addition, we incorporate the physical constraints of the material to the optimization such that the result can be printed in practice. Our simulation experiments demonstrate that our approach brings significant visual quality improvement compared to the multilayer and uniform grid-based approaches. We validate our simulations with fabricated prototypes and demonstrate that our pipeline is flexible enough to allow fabrications of both planar and non-planar displays.

Immersive computer graphics systems strive to generate perceptually realistic user experiences. Current-generation virtual reality (VR) displays are successful in accurately rendering many perceptually important effects, including perspective, disparity, motion parallax, and other depth cues. In this article, we introduce ocular parallax rendering, a technology that accurately renders small amounts of gaze-contingent parallax capable of improving depth perception and realism in VR. Ocular parallax describes the small amounts of depth-dependent image shifts on the retina that are created as the eye rotates. The effect occurs because the centers of rotation and projection of the eye are not the same. We study the perceptual implications of ocular parallax rendering by designing and conducting a series of user experiments. Specifically, we estimate perceptual detection and discrimination thresholds for this effect and demonstrate that it is clearly visible in most VR applications. Additionally, we show that ocular parallax rendering provides an effective ordinal depth cue and it improves the impression of realistic depth in VR.

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.

Holographic displays promise unprecedented capabilities for direct-view displays as well as virtual and augmented reality applications. However, one of the biggest challenges for computer-generated holography (CGH) is the fundamental tradeoff between algorithm runtime and achieved image quality, which has prevented high-quality holographic image synthesis at fast speeds. Moreover, the image quality achieved by most holographic displays is low, due to the mismatch between the optical wave propagation of the display and its simulated model. Here, we develop an algorithmic CGH framework that achieves unprecedented image fidelity and real-time framerates. Our framework comprises several parts, including a novel camera-in-the-loop optimization strategy that allows us to either optimize a hologram directly or train an interpretable model of the optical wave propagation and a neural network architecture that represents the first CGH algorithm capable of generating full-color high-quality holographic images at 1080p resolution in real time.

Occlusion is a powerful visual cue that is crucial for depth perception and realism in optical see-through augmented reality (OST-AR). However, existing OST-AR systems additively overlay physical and digital content with beam combiners - an approach that does not easily support mutual occlusion, resulting in virtual objects that appear semi-transparent and unrealistic. In this work, we propose a new type of occlusion-capable OST-AR system. Rather than additively combining the real and virtual worlds, we employ a single digital micromirror device (DMD) to merge the respective light paths in a multiplicative manner. This unique approach allows us to simultaneously block light incident from the physical scene on a pixel-by-pixel basis while also modulating the light emitted by a light-emitting diode (LED) to display digital content. Our technique builds on mixed binary/continuous factorization algorithms to optimize time-multiplexed binary DMD patterns and their corresponding LED colors to approximate a target augmented reality (AR) scene. In simulations and with a prototype benchtop display, we demonstrate hard-edge occlusions, plausible shadows, and also gaze-contingent optimization of this novel display mode, which only requires a single spatial light modulator.

Feature detectors and descriptors are key low-level vision tools that many higher-level tasks build on. Unfortunately these fail in the presence of challenging light transport effects including partial occlusion, low contrast, and reflective or refractive surfaces. Building on spatio-angular imaging modalities offered by emerging light field cameras, we introduce a new and computationally efficient 4D light field feature detector and descriptor: LiFF. LiFF is scale invariant and utilizes the full 4D light field to detect features that are robust to changes in perspective. This is particularly useful for structure from motion (SfM) and other tasks that match features across viewpoints of a scene. We demonstrate significantly improved 3D reconstructions via SfM when using LiFF instead of the leading 2D or 4D features, and show that LiFF runs an order of magnitude faster than the leading 4D approach. Finally, LiFF inherently estimates depth for each feature, opening a path for future research in light field-based SfM.

Non-line-of-sight (NLOS) imaging enables unprecedented capabilities in a wide range of applications, including robotic and machine vision, remote sensing, autonomous vehicle navigation, and medical imaging. Recent approaches to solving this challenging problem employ optical time-of-flight imaging systems with highly sensitive time-resolved photodetectors and ultra-fast pulsed lasers. However, despite recent successes in NLOS imaging using these systems, widespread implementation and adoption of the technology remains a challenge because of the requirement for specialized, expensive hardware. We introduce acoustic NLOS imaging, which is orders of magnitude less expensive than most optical systems and captures hidden 3D geometry at longer ranges with shorter acquisition times compared to state-of-the-art optical methods. Inspired by hardware setups used in radar and algorithmic approaches to model and invert wave-based image formation models developed in the seismic imaging community, we demonstrate a new approach to seeing around corners.

In this work, we address the lack of 3D understanding of generative neural networks by introducing a persistent 3D feature embedding for view synthesis. To this end, we propose DeepVoxels, a learned representation that encodes the view-dependent appearance of a 3D scene without having to explicitly model its geometry. At its core, our approach is based on a Cartesian 3D grid of persistent embedded features that learn to make use of the underlying 3D scene structure. Our approach combines insights from 3D geometric computer vision with recent advances in learning image-to-image mappings based on adversarial loss functions. DeepVoxels is supervised, without requiring a 3D reconstruction of the scene, using a 2D re-rendering loss and enforces perspective and multi-view geometry in a principled manner. We apply our persistent 3D scene representation to the problem of novel view synthesis demonstrating high-quality results for a variety of challenging scenes.

Depth estimation and 3D object detection are critical for scene understanding but remain challenging to perform with a single image due to the loss of 3D information during image capture. Recent models using deep neural networks have improved monocular depth estimation performance, but there is still difficulty in predicting absolute depth and generalizing outside a standard dataset. Here we introduce the paradigm of deep optics, i.e. end-to-end design of optics and image processing, to the monocular depth estimation problem, using coded defocus blur as an additional depth cue to be decoded by a neural network. We evaluate several optical coding strategies along with an end-to-end optimization scheme for depth estimation on three datasets, including NYU Depth v2 and KITTI. We find an optimized freeform lens design yields the best results, but chromatic aberration from a singlet lens offers significantly improved performance as well. We build a physical prototype and validate that chromatic aberrations improve depth estimation on real-world results. In addition, we train object detection networks on the KITTI dataset and show that the lens optimized for depth estimation also results in improved 3D object detection performance.

Unsupervised learning with generative models has the potential of discovering rich representations of 3D scenes. While geometric deep learning has explored 3D-structure-aware representations of scene geometry, these models typically require explicit 3D supervision. Emerging neural scene representations can be trained only with posed 2D images, but existing methods ignore the three-dimensional structure of scenes. We propose Scene Representation Networks (SRNs), a continuous, 3D-structure-aware scene representation that encodes both geometry and appearance. SRNs represent scenes as continuous functions that map world coordinates to a feature representation of local scene properties. By formulating the image formation as a differentiable ray-marching algorithm, SRNs can be trained end-to-end from only 2D images and their camera poses, without access to depth or shape. This formulation naturally generalizes across scenes, learning powerful geometry and appearance priors in the process. We demonstrate the potential of SRNs by evaluating them for novel view synthesis, few-shot reconstruction, joint shape and appearance interpolation, and unsupervised discovery of a non-rigid face model.

Imaging objects obscured by occluders is a significant challenge for many applications. A camera that could “see around corners” could help improve navigation and mapping capabilities of autonomous vehicles or make search and rescue missions more effective. Time-resolved single-photon imaging systems have recently been demonstrated to record optical information of a scene that can lead to an estimation of the shape and reflectance of objects hidden from the line of sight of a camera. However, existing non-line-of-sight (NLOS) reconstruction algorithms have been constrained in the types of light transport effects they model for the hidden scene parts. We introduce a factored NLOS light transport representation that accounts for partial occlusions and surface normals. Based on this model, we develop a factorization approach for inverse time-resolved light transport and demonstrate high-fidelity NLOS reconstructions for challenging scenes both in simulation and with an experimental NLOS imaging system.

Imaging objects outside a camera's direct line of sight has important applications in robotic vision, remote sensing, and many other domains. Time-of-flight-based non-line-of-sight (NLOS) imaging systems have recently demonstrated impressive results, but several challenges remain. Image formation and inversion models have been slow or limited by the types of hidden surfaces that can be imaged. Moreover, non-planar sampling surfaces and non-confocal scanning methods have not been supported by efficient NLOS algorithms. With this work, we introduce a wave-based image formation model for the problem of NLOS imaging. Inspired by inverse methods used in seismology, we adapt a frequency-domain method, f-k migration, for solving the inverse NLOS problem. Unlike existing NLOS algorithms, f-k migration is both fast and memory efficient, it is robust to specular and other complex reflectance properties, and we show how it can be used with non-confocally scanned measurements as well as for non-planar sampling surfaces. f-k migration is more robust to measurement noise than alternative methods, generally produces better quality reconstructions, and is easy to implement. We experimentally validate our algorithms with a new NLOS imaging system that records room-sized scenes outdoors under indirect sunlight, and scans persons wearing retroreflective clothing at interactive rates.

Holographic near-eye displays are a key enabling technology for virtual and augmented reality (VR/AR) applications. Holographic stereograms (HS) are a method of encoding a light field into a hologram, which enables them to natively support view-dependent lighting effects. However, existing HS algorithms require the choice of a hogel size, forcing a tradeoff between spatial and angular resolution. Based on the fact that the short-time Fourier transform (STFT) connects a hologram to its observable light field, we develop the overlap-add stereogram (OLAS) as the correct method of "inverting" the light field into a hologram via the STFT. The OLAS makes more efficient use of the information contained within the light field than previous HS algorithms, exhibiting better image quality at a range of distances and hogel sizes. Most remarkably, the OLAS does not degrade spatial resolution with increasing hogel size, overcoming the spatio-angular resolution tradeoff that previous HS algorithms face. Importantly, the optimal hogel size of previous methods typically varies with the depth of every object in a scene, making the OLAS not only a hogel size-invariant method, but also nearly scene independent. We demonstrate the performance of the OLAS both in simulation and on a prototype near-eye display system, showing focusing capabilities and view-dependent effects.

Typical camera optics consist of a system of individual elements that are designed to compensate for the aberrations of a single lens. Recent computational cameras shift some of this correction task from the optics to post-capture processing, reducing the imaging optics to only a few optical elements. However, these systems only achieve reasonable image quality by limiting the field of view (FOV) to a few degrees - effectively ignoring severe off-axis aberrations with blur sizes of multiple hundred pixels. In this paper, we propose a lens design and learned reconstruction architecture that lift this limitation and provide an order of magnitude increase in field of view using only a single thin-plate lens element. Specifically, we design a lens to produce spatially shift-invariant point spread functions, over the full FOV, that are tailored to the proposed reconstruction architecture. We achieve this with a mixture PSF, consisting of a peak and and a low-pass component, which provides residual contrast instead of a small spot size as in traditional lens designs. To perform the reconstruction, we train a deep network on captured data from a display lab setup, eliminating the need for manual acquisition of training data in the field. We assess the proposed method in simulation and experimentally with a prototype camera system. We compare our system against existing single-element designs, including an aspherical lens and a pinhole, and we compare against a complex multielement lens, validating high-quality large field-of-view (i.e. 53°) imaging performance using only a single thin-plate element.

Optical see-through augmented reality (AR) systems are a next-generation computing platform that offer unprecedented user experiences by seamlessly combining physical and digital content. Many of the traditional challenges of these displays have been significantly improved over the last few years, but AR experiences offered by today's systems are far from seamless and perceptually realistic. Mutually consistent occlusions between physical and digital objects are typically not supported. When mutual occlusion is supported, it is only supported for a fixed depth. We propose a new optical see-through AR display system that renders mutual occlusion in a depth-dependent, perceptually realistic manner. To this end, we introduce varifocal occlusion displays based on focus-tunable optics, which comprise a varifocal lens system and spatial light modulators that enable depth-corrected hard-edge occlusions for AR experiences. We derive formal optimization methods and closed-form solutions for driving this tunable lens system and demonstrate a monocular varifocal occlusion-capable optical see-through AR display capable of perceptually realistic occlusion across a large depth range.

We present an end-to-end image processing framework for time-of-flight (ToF) cameras. Existing ToF image processing pipelines consist of a sequence of operations including modulated exposures, denoising, phase unwrapping and multipath interference correction. While this cascaded modular design offers several benefits, such as closed-form solutions and power-efficient processing, it also suffers from error accumulation and information loss as each module can only observe the output from its direct predecessor, resulting in erroneous depth estimates. We depart from a conventional pipeline model and propose a deep convolutional neural network architecture that recovers scene depth directly from dual-frequency, raw ToF correlation measurements. To train this network, we simulate ToF images for a variety of scenes using a time-resolved renderer, devise depth-specific losses, and apply normalization and augmentation strategies to generalize this model to real captures. We demonstrate that the proposed network can efficiently exploit the spatio-temporal structures of ToF frequency measurements, and validate the performance of the joint multipath removal, denoising and phase unwrapping method on a wide range of challenging scenes.

Active imaging at the picosecond timescale reveals transient light transport effects otherwise not accessible by computer vision and image processing algorithms. For example, analyzing the time of flight of short laser pulses emitted into a scene and scattered back to a detector allows for depth imaging, which is crucial for autonomous driving and many other applications. Moreover, analyzing or removing global light transport effects from photographs becomes feasible. While several transient imaging systems have recently been proposed using various imaging technologies, none is capable of acquiring transient images at interactive framerates. In this paper, we present an imaging system that records transient images at up to 25 Hz. We show several transient video clips recorded with this system and demonstrate transient imaging applications, including direct-global light transport separation and enhanced depth imaging.

Emerging sensor designs increasingly rely on novel color filter arrays (CFAs) to sample the incident spectrum in unconventional ways. In particular, capturing a near-infrared (NIR) channel along with conventional RGB color is an exciting new imaging modality. RGB+NIR sensing has broad applications in computational photography, such as low-light denoising, it has applications in computer vision, such as facial recognition and tracking, and it paves the way toward low-cost single-sensor RGB and depth imaging using structured illumination. However, cost-effective commercial CFAs suffer from severe spectral cross talk. This cross talk represents a major challenge in high-quality RGB+NIR imaging, rendering existing spatially multiplexed sensor designs impractical. In this work, we introduce a new approach to RGB+NIR image reconstruction using learned convolutional sparse priors. We demonstrate high-quality color and NIR imaging for challenging scenes, even including high-frequency structured NIR illumination. The effectiveness of the proposed method is validated on a large data set of experimental captures, and simulated benchmark results which demonstrate that this work achieves unprecedented reconstruction quality.

Sensors which capture 3D scene information provide useful data for tasks in vehicle navigation, gesture recognition, human pose estimation, and geometric reconstruction. Active illumination time-of-flight sensors in particular have become widely used to estimate a 3D representation of a scene. However, the maximum range, density of acquired spatial samples, and overall acquisition time of these sensors is fundamentally limited by the minimum signal required to estimate depth reliably. In this paper, we propose a data-driven method for photon-efficient 3D imaging which leverages sensor fusion and computational reconstruction to rapidly and robustly estimate a dense depth map from low photon counts. Our sensor fusion approach uses measurements of single photon arrival times from a low-resolution single-photon detector array and an intensity image from a conventional high-resolution camera. Using a multi-scale deep convolutional network, we jointly process the raw measurements from both sensors and output a high-resolution depth map. To demonstrate the efficacy of our approach, we implement a hardware prototype and show results using captured data. At low signal-to-background levels, our depth reconstruction algorithm with sensor fusion outperforms other methods for depth estimation from noisy measurements of photon arrival times.

In typical cameras the optical system is designed first; once it is fixed, the parameters in the image processing algorithm are tuned to get good image reproduction. In contrast to this sequential design approach, we consider joint optimization of an optical system (for example, the physical shape of the lens) together with the parameters of the reconstruction algorithm. We build a fully-differentiable simulation model that maps the true source image to the reconstructed one. The model includes diffractive light propagation, depth and wavelength-dependent effects, noise and nonlinearities, and the image post-processing. We jointly optimize the optical parameters and the image processing algorithm parameters so as to minimize the deviation between the true and reconstructed image, over a large set of images. We implement our joint optimization method using autodifferentiation to efficiently compute parameter gradients in a stochastic optimization algorithm. We demonstrate the efficacy of this approach by applying it to achromatic extended depth of field and snapshot super-resolution imaging.

Virtual reality systems are widely believed to be the next major computing platform. There are, however, some barriers to adoption that must be addressed, such as that of motion sickness - which can lead to undesirable symptoms including postural instability, headaches, and nausea. Motion sickness in virtual reality occurs as a result of moving visual stimuli that cause users to perceive self-motion while they remain stationary in the real world. There are several contributing factors to both this perception of motion and the subsequent onset of sickness, including field of view, motion velocity, and stimulus depth. We verify first that differences in vection due to relative stimulus depth remain correlated with sickness. Then, we build a dataset of stereoscopic 3D videos and their corresponding sickness ratings in order to quantify their nauseogenicity, which we make available for future use. Using this dataset, we train a machine learning algorithm on hand-crafted features (quantifying speed, direction, and depth as functions of time) from each video, learning the contributions of these various features to the sickness ratings. Our predictor generally outperforms a naïve estimate, but is ultimately limited by the size of the dataset. However, our result is promising and opens the door to future work with more extensive datasets. This and further advances in this space have the potential to alleviate developer and end user concerns about motion sickness in the increasingly commonplace virtual world.

Understanding how people explore immersive virtual environments is crucial for many applications, such as designing virtual reality (VR) content, developing new compression algorithms, or learning computational models of saliency or visual attention. Whereas a body of recent work has focused on modeling saliency in desktop viewing conditions, VR is very different from these conditions in that viewing behavior is governed by stereoscopic vision and by the complex interaction of head orientation, gaze, and other kinematic constraints. To further our understanding of viewing behavior and saliency in VR, we capture and analyze gaze and head orientation data of 169 users exploring stereoscopic, static omni-directional panoramas, for a total of 1980 head and gaze trajectories for three different viewing conditions. We provide a thorough analysis of our data, which leads to several important insights, such as the existence of a particular fixation bias, which we then use to adapt existing saliency predictors to immersive VR conditions. In addition, we explore other applications of our data and analysis, including automatic alignment of VR video cuts, panorama thumbnails, panorama video synopsis, and saliency-basedcompression.

Light field (LF) capture and processing are important in an expanding range of computer vision applications, offering rich textural and depth information and simplification of conventionally complex tasks. Although LF cameras are commercially available, no existing device offers wide field-of-view (FOV) imaging. This is due in part to the limitations of fisheye lenses, for which a fundamentally constrained entrance pupil diameter severely limits depth sensitivity. In this work we describe a novel, compact optical design that couples a monocentric lens with multiple sensors using microlens arrays, allowing LF capture with an unprecedented FOV. Leveraging capabilities of the LF representation, we propose a novel method for efficiently coupling the spherical lens and planar sensors, replacing expensive and bulky fiber bundles. We construct a single-sensor LF camera prototype, rotating the sensor relative to a fixed main lens to emulate a wide-FOV multi-sensor scenario. Finally, we describe a processing toolchain, including a convenient spherical LF parameterization, and demonstrate depth estimation and post-capture refocus for indoor and outdoor panoramas with 15 × 15 × 1600 × 200 pixels (72 MPix) and a 138° FOV.

Computer vision algorithms build on 2D images or 3D videos that capture dynamic events at the millisecond time scale. However, capturing and analyzing “transient images” at the picosecond scale-i.e., at one trillion frames per second-reveals unprecedented information about a scene and light transport within. This is not only crucial for time-of-flight range imaging, but it also helps further our understanding of light transport phenomena at a more fundamental level and potentially allows to revisit many assumptions made in different computer vision algorithms. In this work, we design and evaluate an imaging system that builds on single photon avalanche diode (SPAD) sensors to capture multi-path responses with picosecond-scale active illumination. We develop inverse methods that use modern approaches to deconvolve and denoise measurements in the presence of Poisson noise, and compute transient images at a higher quality than previously reported. The small form factor, fast acquisition rates, and relatively low cost of our system potentially makes transient imaging more practical for a range of applications.

Light field microscopy (LFM) is an emerging technique for volumetric fluorescence imaging, but widespread use is hampered by its poor spatial resolution. Using diffraction-based analysis we show how this degraded resolution arises because conventional LFM aims to sample four dimensions of the light field. By instead prioritizing 3D volumetric information over 4D sampling, we can optically interfere certain redundant angular samples to allow higher spatial resolution while maintaining enough angular information for depth discrimination. With this in mind, we design a number of aperture plane sampling schemes, characterize their frequency support and invertibility, and describe how their relative performance depends on the operating signal-to-noise regime. With simulations and a prototype, we demonstrate a time-sequential amplitude mask-based acquisition approach that outperforms conventional LFM in terms of both spatial resolution and axial field of view.

Computational photography encompasses a diversity of imaging techniques, but one of the core operations performed by many of them is to compute image differences. An intuitive approach to computing such differences is to capture several images sequentially and then process them jointly. In this paper, we introduce a snapshot difference imaging approach that is directly implemented in the sensor hardware of emerging time-of-flight cameras. With a variety of examples, we demonstrate that the proposed snapshot difference imaging technique is useful for direct-global illumination separation, for direct imaging of spatial and temporal image gradients, for direct depth edge imaging, and more.

Streaming of 360° content is gaining attention as an immersive way to remotely experience live events. However live capture is presently limited to 2D content due to the prohibitive computational cost associated with multi-camera rigs. In this work we present a system that directly captures streaming 3D virtual reality content. Our approach does not suffer from spatial or temporal seams and natively handles phenomena that are challenging for existing systems, including refraction, reflection, transparency and speculars. Vortex natively captures in the omni-directional stereo (ODS) format, which is widely supported by VR displays and streaming pipelines. We identify an important source of distortion inherent to the ODS format, and demonstrate a simple means of correcting it. We include a detailed analysis of the design space, including tradeoffs between noise, frame rate, resolution, and hardware complexity. Processing is minimal, enabling live transmission of immersive, 3D, 360° content. We construct a prototype and demonstrate capture of 360° scenes at up to 8192 X 4096 pixels at 5 fps, and establish the viability of operation up to 32 fps.

Although emerging virtual and augmented reality (VR/AR) systems can produce highly immersive experiences, they can also cause visual discomfort, eyestrain, and nausea. One of the sources of these symptoms is a mismatch between vergence and focus cues. In current VR/AR near-eye displays, a stereoscopic image pair drives the vergence state of the human visual system to arbitrary distances, but the accommodation, or focus, state of the eyes is optically driven towards a fixed distance. In this work, we introduce a new display technology, dubbed accommodation-invariant (AI) near-eye displays, to improve the consistency of depth cues in near-eye displays. Rather than producing correct focus cues, AI displays are optically engineered to produce visual stimuli that are invariant to the accommodation state of the eye. The accommodation system can then be driven by stereoscopic cues, and the mismatch between vergence and accommodation state of the eyes is significantly reduced. We validate the principle of operation of AI displays using a prototype display that allows for the accommodation state of users to be measured while they view visual stimuli using multiple different display modes.

Traditional cinematography has relied for over a century on a well-established set of editing rules, called continuity editing, to create a sense of situational continuity. Despite massive changes in visual content across cuts, viewers in general experience no trouble perceiving the discontinuous flow of information as a coherent set of events. However, Virtual Reality (VR) movies are intrinsically different from traditional movies in that the viewer controls the camera orientation at all times. As a consequence, common editing techniques that rely on camera orientations, zooms, etc., cannot be used. In this paper we investigate key relevant questions to understand how well traditional movie editing carries over to VR, such as: Does the perception of continuity hold across edit boundaries? Under which conditions? Does viewers' observational behavior change after the cuts? To do so, we rely on recent cognition studies and the event segmentation theory, which states that our brains segment continuous actions into a series of discrete, meaningful events. We first replicate one of these studies to assess whether the predictions of such theory can be applied to VR. We next gather gaze data from viewers watching VR videos containing different edits with varying parameters, and provide the first systematic analysis of viewers' behavior and the perception of continuity in VR. From this analysis we make a series of relevant findings; for instance, our data suggests that predictions from the cognitive event segmentation theory are useful guides for VR editing; that different types of edits are equally well understood in terms of continuity; and that spatial misalignments between regions of interest at the edit boundaries favor a more exploratory behavior even after viewers have fixated on a new region of interest. In addition, we propose a number of metrics to describe viewers' attentional behavior in VR. We believe the insights derived from our work can be useful as guidelines for VR content creation.

Emerging virtual reality (VR) displays must overcome the prevalent issue of visual discomfort to provide high-quality and immersive user experiences. In particular, the mismatch between vergence and accommodation cues inherent to most stereoscopic displays has been a long standing challenge. In this paper, we evaluate several adaptive display modes afforded by focus-tunable optics or actuated displays that have the promise to mitigate visual discomfort caused by the vergence-accommodation conflict, and improve performance in VR environments. We also explore monovision as an unconventional mode that allows each eye of an observer to accommodate to a different distance. While this technique is common practice in ophthalmology, we are the first to report its effectiveness for VR applications with a custom built set up. We demonstrate that monovision and other focus-tunable display modes can provide better user experiences and improve user performance in terms of reaction times and accuracy, particularly for nearby simulated distances in VR.

Light fields have many applications in machine vision, consumer photography, robotics, and microscopy. However, the prevalent resolution limits of existing light field imaging systems hinder widespread adoption. In this paper, we analyze fundamental resolution limits of light field cameras in the diffraction limit. We propose a sequential, coded-aperture-style acquisition scheme that optimizes the resolution of a light field reconstructed from multiple photographs captured from different perspectives and fnumber settings. We also show that the proposed acquisition scheme facilitates high dynamic range light field imaging and demonstrate a proof-of-concept prototype system. With this work, we hope to advance our understanding of the resolution limits of light field photography and develop practical computational imaging systems to overcome them.

Cinematic virtual reality (VR) aims to provide immersive visual experiences of real-world scenes on head-mounted displays. Current cinematic VR systems support stereo cues, but not focus cues that are important for depth perception and comfortable viewing. We propose a new content representation, depth augmented stereo panorama (DASP), which permits generating light fields (LFs) across the observer's pupils, achieving an order of magnitude reduction in data requirements compared to the existing techniques. The LF generation and refocusing capabilities of DASP are evaluated for both, computer-generated and real scenes. Results indicate that DASP can successfully create stereo as well as support focus cues.

Abstract Current HDR acquisition techniques are based on either (i) fusing multibracketed, low dynamic range (LDR) images, (ii) modifying existing hardware and capturing different exposures simultaneously with multiple sensors, or (iii) reconstructing a single image with spatially‐varying pixel exposures. In this paper, we propose a novel algorithm to recover high‐quality HDRI images from a single, coded exposure. The proposed reconstruction method builds on recently‐introduced ideas of convolutional sparse coding (CSC); this paper demonstrates how to make CSC practical for HDR imaging. We demonstrate that the proposed algorithm achieves higher‐quality reconstructions than alternative methods, we evaluate optical coding schemes, analyze algorithmic parameters, and build a prototype coded HDR camera that demonstrates the utility of convolutional sparse HDRI coding with a custom hardware platform.

Computational photography systems are becoming increasingly diverse, while computational resources---for example on mobile platforms---are rapidly increasing. As diverse as these camera systems may be, slightly different variants of the underlying image processing tasks, such as demosaicking, deconvolution, denoising, inpainting, image fusion, and alignment, are shared between all of these systems. Formal optimization methods have recently been demonstrated to achieve state-of-the-art quality for many of these applications. Unfortunately, different combinations of natural image priors and optimization algorithms may be optimal for different problems, and implementing and testing each combination is currently a time-consuming and error-prone process. ProxImaL is a domain-specific language and compiler for image optimization problems that makes it easy to experiment with different problem formulations and algorithm choices. The language uses proximal operators as the fundamental building blocks of a variety of linear and nonlinear image formation models and cost functions, advanced image priors, and noise models. The compiler intelligently chooses the best way to translate a problem formulation and choice of optimization algorithm into an efficient solver implementation. In applications to the image processing pipeline, deconvolution in the presence of Poisson-distributed shot noise, and burst denoising, we show that a few lines of ProxImaL code can generate highly efficient solvers that achieve state-of-the-art results. We also show applications to the nonlinear and nonconvex problem of phase retrieval.

Depth cameras are a ubiquitous technology used in a wide range of applications, including robotic and machine vision, human-computer interaction, autonomous vehicles as well as augmented and virtual reality. In this paper, we explore the design and applications of phased multi-camera time-of-flight (ToF) systems. We develop a reproducible hardware system that allows for the exposure times and waveforms of up to three cameras to be synchronized. Using this system, we analyze waveform interference between multiple light sources in ToF applications and propose simple solutions to this problem. Building on the concept of orthogonal frequency design, we demonstrate state-of-the-art results for instantaneous radial velocity capture via Doppler time-of-flight imaging and we explore new directions for optically probing global illumination, for example by de-scattering dynamic scenes and by non-line-of-sight motion detection via frequency gating.

Convolutional sparse coding (CSC) has become an increasingly important tool in machine learning and computer vision. Image features can be learned and subsequently used for classification and reconstruction tasks. As opposed to patch-based methods, convolutional sparse coding operates on whole images, thereby seamlessly capturing the correlation between local neighborhoods. In this paper, we propose a new approach to solving CSC problems and show that our method converges significantly faster and also finds better solutions than the state of the art. In addition, the proposed method is the first efficient approach to allow for proper boundary conditions to be imposed and it also supports feature learning from incomplete data as well as general reconstruction problems.

Over the last few years, depth cameras have become increasingly popular for a range of applications, including human-computer interaction and gaming, augmented reality, machine vision, and medical imaging. Many of the commercially-available devices use the time-of-flight principle, where active illumination is temporally coded and analyzed in the camera to estimate a per-pixel depth map of the scene. In this paper, we propose a fundamentally new imaging modality for all time-of-flight (ToF) cameras: per-pixel radial velocity measurement. The proposed technique exploits the Doppler effect of objects in motion, which shifts the temporal illumination frequency before it reaches the camera. Using carefully coded illumination and modulation frequencies of the ToF camera, object velocities directly map to measured pixel intensities. We show that a slight modification of our imaging system allows for color, depth, and velocity information to be captured simultaneously. Combining the optical flow computed on the RGB frames with the measured metric radial velocity allows us to further estimate the full 3D metric velocity field of the scene. The proposed technique has applications in many computer graphics and vision problems, for example motion tracking, segmentation, recognition, and motion deblurring.

Over the last few years, virtual reality (VR) has re-emerged as a technology that is now feasible at low cost via inexpensive cellphone components. In particular, advances of high-resolution micro displays, low-latency orientation trackers, and modern GPUs facilitate immersive experiences at low cost. One of the remaining challenges to further improve visual comfort in VR experiences is the vergence-accommodation conflict inherent to all stereoscopic displays. Accurate reproduction of all depth cues is crucial for visual comfort. By combining well-known stereoscopic display principles with emerging factored light field technology, we present the first wearable VR display supporting high image resolution as well as focus cues. A light field is presented to each eye, which provides more natural viewing experiences than conventional near-eye displays. Since the eye box is just slightly larger than the pupil size, rank-1 light field factorizations are sufficient to produce correct or nearly-correct focus cues; no time-multiplexed image display or gaze tracking is required. We analyze lens distortions in 4D light field space and correct them using the afforded high-dimensional image formation. We also demonstrate significant improvements in resolution and retinal blur quality over related near-eye displays. Finally, we analyze diffraction limits of these types of displays.

Fundamental display characteristics are constantly being improved, especially resolution, dynamic range, and color reproduction. However, whereas high resolution and high-dynamic range displays have matured as a technology, it remains largely unclear how to extend the color gamut of a display without either sacrificing light throughput or making other tradeoffs. In this paper, we advocate for adaptive color display; with hardware implementations that allow for color primaries to be dynamically chosen, an optimal gamut and corresponding pixel states can be computed in a content-adaptive and user-centric manner. We build a flexible gamut projector and develop a perceptually-driven optimization framework that robustly factors a wide color gamut target image into a set of time-multiplexed primaries and corresponding pixel values. We demonstrate that adaptive primary selection has many benefits over fixed gamut selection and show that our algorithm for joint primary selection and gamut mapping performs better than existing methods. Finally, we evaluate the proposed computational display system extensively in simulation and, via photographs and user experiments, with a prototype adaptive color projector.

Capturing and understanding visual signals is one of the core interests of computer vision. Much progress has been made w.r.t. many aspects of imaging, but the reconstruc-tion of refractive phenomena, such as turbulence, gas and heat flows, liquids, or transparent solids, has remained a challenging problem. In this paper, we derive an intuitive formulation of light transport in refractive media using light fields and the transport of intensity equation. We show how coded illumination in combination with pairs of recorded images allow for robust computational reconstruction of dy-namic two and three-dimensional refractive phenomena. 1.

We propose a flexible light field camera architecture that is at the convergence of optics, sensor electronics, and applied mathematics. Through the co-design of a sensor that comprises tailored, Angle Sensitive Pixels and advanced reconstruction algorithms, we show that—contrary to light field cameras today—our system can use the same measurements captured in a single sensor image to recover either a high-resolution 2D image, a low-resolution 4D light field using fast, linear processing, or a high-resolution light field using sparsity-constrained optimization.

For about a century, researchers and experimentalists have strived to bring glasses-free 3D experiences to the big screen. Much progress has been made and light field projection systems are now commercially available. Unfortunately, available display systems usually employ dozens of devices making such setups costly, energy inefficient, and bulky. We present a compressive approach to light field synthesis with projection devices. For this purpose, we propose a novel, passive screen design that is inspired by angle-expanding Keplerian telescopes. Combined with high-speed light field projection and nonnegative light field factorization, we demonstrate that compressive light field projection is possible with a single device. We build a prototype light field projector and angle-expanding screen from scratch, evaluate the system in simulation, present a variety of results, and demonstrate that the projector can alternatively achieve super-resolved and high dynamic range 2D image display when used with a conventional screen.

Millions of people worldwide need glasses or contact lenses to see or read properly. We introduce a computational display technology that predistorts the presented content for an observer, so that the target image is perceived without the need for eyewear. By designing optics in concert with prefiltering algorithms, the proposed display architecture achieves significantly higher resolution and contrast than prior approaches to vision-correcting image display. We demonstrate that inexpensive light field displays driven by efficient implementations of 4D prefiltering algorithms can produce the desired vision-corrected imagery, even for higher-order aberrations that are difficult to be corrected with glasses. The proposed computational display architecture is evaluated in simulation and with a low-cost prototype device.

We present coded focal stack photography as a computational photography paradigm that combines a focal sweep and a coded sensor readout with novel computational algorithms. We demonstrate various applications of coded focal stacks, including photography with programmable non-planar focal surfaces and multiplexed focal stack acquisition. By leveraging sparse coding techniques, coded focal stacks can also be used to recover a full-resolution depth and all-in-focus (AIF) image from a single photograph. Coded focal stack photography is a significant step towards a computational camera architecture that facilitates high-resolution post-capture refocusing, flexible depth of field, and 3D imaging.

Projectors require large apertures to maximize light throughput. Unfortunately, this leads to shallow depths of field (DOF), hence blurry images, when projecting on non-planar surfaces, such as cultural heritage sites, curved screens, or when sharing visual information in everyday environments. We introduce high-rank coded aperture projectors - a new computational display technology that combines optical designs with computational processing to overcome depth of field limitations of conventional devices. In particular, we employ high-speed spatial light modulators (SLMs) on the image plane and in the aperture of modified projectors. The patterns displayed on these SLMs are computed with a new mathematical framework that uses high-rank light field factorizations and directly exploits the limited temporal resolution and contrast sensitivity of the human visual system. With an experimental prototype projector, we demonstrate significantly increased DOF as compared to conventional technology.

Recent years have seen proposals for exciting new computational display technologies that arecompressivein the sense that they generate high resolution images or light fields with relatively few display parameters. Image synthesis for these types of displays involves two major tasks: sampling and rendering high-dimensional target imagery, such as light fields or time-varying light fields, as well as optimizing the display parameters to provide a good approximation of the target content. In this paper, we introduce an adaptive optimization framework for compressive displays that generates high quality images and light fields using only a fraction of the total plenoptic samples. We demonstrate the framework for a large set of display technologies, including several types of auto-stereoscopic displays, high dynamic range displays, and high-resolution displays. We achieve significant performance gains, and in some cases are able to process data that would be infeasible with existing methods.

We present a glasses-free 3D display design with the potential to provide viewers with nearly correct accommodative depth cues, as well as motion parallax and binocular cues. Building on multilayer attenuator and directional backlight architectures, the proposed design achieves the high angular resolution needed for accommodation by placing spatial light modulators about a large lens: one conjugate to the viewer's eye, and one or more near the plane of the lens. Nonnegative tensor factorization is used to compress a high angular resolution light field into a set of masks that can be displayed on a pair of commodity LCD panels. By constraining the tensor factorization to preserve only those light rays seen by the viewer, we effectively steer narrow high-resolution viewing cones into the user's eyes, allowing binocular disparity, motion parallax, and the potential for nearly correct accommodation over a wide field of view. We verify the design experimentally by focusing a camera at different depths about a prototype display, establish formal upper bounds on the design's accommodation range and diffraction-limited performance, and discuss practical limitations that must be overcome to allow the device to be used with human observers.

Light field photography has gained a significant research interest in the last two decades; today, commercial light field cameras are widely available. Nevertheless, most existing acquisition approaches either multiplex a low-resolution light field into a single 2D sensor image or require multiple photographs to be taken for acquiring a high-resolution light field. We propose a compressive light field camera architecture that allows for higher-resolution light fields to be recovered than previously possible from a single image. The proposed architecture comprises three key components: light field atoms as a sparse representation of natural light fields, an optical design that allows for capturing optimized 2D light field projections, and robust sparse reconstruction methods to recover a 4D light field from a single coded 2D projection. In addition, we demonstrate a variety of other applications for light field atoms and sparse coding, including 4D light field compression and denoising.

We introduce tensor displays: a family of compressive light field displays comprising all architectures employing a stack of time-multiplexed, light-attenuating layers illuminated by uniform or directional backlighting (i.e., any low-resolution light field emitter). We show that the light field emitted by an N -layer, M -frame tensor display can be represented by an N th -order, rank- M tensor. Using this representation we introduce a unified optimization framework, based on nonnegative tensor factorization (NTF), encompassing all tensor display architectures. This framework is the first to allow joint multilayer, multiframe light field decompositions, significantly reducing artifacts observed with prior multilayer-only and multiframe-only decompositions; it is also the first optimization method for designs combining multiple layers with directional backlighting. We verify the benefits and limitations of tensor displays by constructing a prototype using modified LCD panels and a custom integral imaging backlight. Our efficient, GPU-based NTF implementation enables interactive applications. Through simulations and experiments we show that tensor displays reveal practical architectures with greater depths of field, wider fields of view, and thinner form factors, compared to prior automultiscopic displays.

We introduce a new approach to capturing refraction in transparent media, which we call Light Field Background Oriented Schlieren Photography (LFBOS). By optically coding the locations and directions of light rays emerging from a light field probe, we can capture changes of the refractive index field between the probe and a camera or an observer. Rather than using complicated and expensive optical setups as in traditional Schlieren photography we employ commodity hardware; our prototype consists of a camera and a lenslet array. By carefully encoding the color and intensity variations of a 4D probe instead of a diffuse 2D background, we avoid expensive computational processing of the captured data, which is necessary for Background Oriented Schlieren imaging (BOS). We analyze the benefits and limitations of our approach and discuss application scenarios.

Acquiring transparent, refractive objects is challenging as these kinds of objects can only be observed by analyzing the distortion of reference background patterns. We present a new, single image approach to reconstructing thin transparent surfaces, such as thin solids or surfaces of fluids. Our method is based on observing the distortion of light field background illumination. Light field probes have the potential to encode up to four dimensions in varying colors and intensities: spatial and angular variation on the probe surface; commonly employed reference patterns are only two-dimensional by coding either position or angle on the probe. We show that the additional information can be used to reconstruct refractive surface normals and a sparse set of control points from a single photograph.

Abstract The plenoptic function is a ray‐based model for light that includes the colour spectrum as well as spatial, temporal and directional variation. Although digital light sensors have greatly evolved in the last years, one fundamental limitation remains: all standard CCD and CMOS sensors integrate over the dimensions of the plenoptic function as they convert photons into electrons; in the process, all visual information is irreversibly lost, except for a two‐dimensional, spatially varying subset—the common photograph. In this state‐of‐the‐art report, we review approaches that optically encode the dimensions of the plenoptic function transcending those captured by traditional photography and reconstruct the recorded information computationally.

We introduce polarization field displays as an optically-efficient design for dynamic light field display using multi-layered LCDs. Such displays consist of a stacked set of liquid crystal panels with a single pair of crossed linear polarizers. Each layer is modeled as a spatially-controllable polarization rotator, as opposed to a conventional spatial light modulator that directly attenuates light. Color display is achieved using field sequential color illumination with monochromatic LCDs, mitigating severe attenuation and moiré occurring with layered color filter arrays. We demonstrate such displays can be controlled, at interactive refresh rates, by adopting the SART algorithm to tomographically solve for the optimal spatially-varying polarization state rotations applied by each layer. We validate our design by constructing a prototype using modified off-the-shelf panels. We demonstrate interactive display using a GPU-based SART implementation supporting both polarization-based and attenuation-based architectures. Experiments characterize the accuracy of our image formation model, verifying polarization field displays achieve increased brightness, higher resolution, and extended depth of field, as compared to existing automultiscopic display methods for dual-layer and multi-layer LCDs.

We develop tomographic techniques for image synthesis on displays composed of compact volumes of light-attenuating material. Such volumetric attenuators recreate a 4D light field or high-contrast 2D image when illuminated by a uniform backlight. Since arbitrary oblique views may be inconsistent with any single attenuator, iterative tomographic reconstruction minimizes the difference between the emitted and target light fields, subject to physical constraints on attenuation. As multi-layer generalizations of conventional parallax barriers, such displays are shown, both by theory and experiment, to exceed the performance of existing dual-layer architectures. For 3D display, spatial resolution, depth of field, and brightness are increased, compared to parallax barriers. For a plane at a fixed depth, our optimization also allows optimal construction of high dynamic range displays, confirming existing heuristics and providing the first extension to multiple, disjoint layers. We conclude by demonstrating the benefits and limitations of attenuation-based light field displays using an inexpensive fabrication method: separating multiple printed transparencies with acrylic sheets.

Multiplexing is a common technique for encoding high-dimensional image data into a single, two-dimensional image. Examples of spatial multiplexing include Bayer patterns to capture color channels, and integral images to encode light fields. In the Fourier domain, optical heterodyning has been used to acquire light fields. In this paper, we develop a general theory of multiplexing the dimensions of the plenoptic function onto an image sensor. Our theory enables a principled comparison of plenoptic multiplexing schemes, including noise analysis, as well as the development of a generic reconstruction algorithm. The framework also aides in the identification and optimization of novel multiplexed imaging applications.

Optically multiplexed image acquisition techniques have become increasingly popular for encoding different exposures, color channels, light fields, and other properties of light onto two-dimensional image sensors. Recently, Fourier-based multiplexing and reconstruction approaches have been introduced in order to achieve a superior light transmission of the employed modulators and better signal-to-noise characteristics of the reconstructed data. We show in this paper that Fourier-based reconstruction approaches suffer from severe artifacts in the case of sensor saturation, i.e. when the dynamic range of the scene exceeds the capabilities of the image sensor. We analyze the problem, and propose a novel combined optical light modulation and computational reconstruction method that not only suppresses such artifacts, but also allows us to recover a wider dynamic range than existing image-space multiplexing approaches.

Abstract We propose to enhance the capabilities of the human visual system by performing optical image processing directly on an observed scene. Unlike previous work which additively superimposes imagery on a scene, or completely replaces scene imagery with a manipulated version, we perform all manipulation through the use of a light modulation display to spatially filter incoming light. We demonstrate a number of perceptually motivated algorithms including contrast enhancement and reduction, object highlighting for preattentive emphasis, colour saturation, de‐saturation and de‐metamerization, as well as visual enhancement for the colour blind. A camera observing the scene guides the algorithms for on‐the‐fly processing, enabling dynamic application scenarios such as monocular scopes, eyeglasses and windshields.

Coding a projector's aperture plane with adaptive patterns together with inverse filtering allow the depth-of-field of projected imagery to be increased. We present two prototypes and corresponding algorithms for static and programmable apertures. We also explain how these patterns can be computed at interactive rates, by taking into account the image content and limitations of the human visual system. Applications such as projector defocus compensation, high-quality projector depixelation, and increased temporal contrast of projected video sequences can be supported. Coded apertures are a step towards next-generation auto-iris projector lenses.

Abstract This article focuses on real‐time image correction techniques that enable projector‐camera systems to display images onto screens that are not optimized for projections, such as geometrically complex, coloured and textured surfaces. It reviews hardware‐accelerated methods like pixel‐precise geometric warping, radiometric compensation, multi‐focal projection and the correction of general light modulation effects. Online and offline calibration as well as invisible coding methods are explained. Novel attempts in super‐resolution, high‐dynamic range and high‐speed projection are discussed. These techniques open a variety of new applications for projection displays. Some of them will also be presented in this report.

Radiometric compensation techniques allow seamless projections onto complex everyday surfaces. Implemented with projector-camera systems they support the presentation of visual content in situations where projection-optimized screens are not available or not desired - as in museums, historic sites, air-plane cabins, or stage performances. We propose a novel approach that employs the full light transport between projectors and a camera to account for many illumination aspects, such as interreflections, refractions, shadows, and defocus. Precomputing the inverse light transport in combination with an efficient implementation on the GPU makes the real-time compensation of captured local and global light modulations possible.

We show how view-dependent image-based and geometric warping, radiometric compensation, and multi-focal projection enable a view-dependent stereoscopic visualization on ordinary (geometrically complex, colored and textured) surfaces within everyday environments. Special display configurations for immersive or semi-immersive AR/VR applications that require permanent and artificial projection canvases might become unnecessary. We demonstrate several ad-hoc visualization examples in a real architectural and museum application context.

We present techniques which create a consistent illumination between real and virtual objects inside an application specific optical see-through display: the virtual showcase. We use projectors and cameras to capture reflectance information from diffuse real objects and to illuminate them under new synthetic lighting conditions. Matching direct and indirect lighting effects, such as shading, shadows, reflections and color bleeding can be approximated at interactive rates in such a controlled mixed environment.

No abstract available.