The recent success of text-to-image synthesis has taken the world by storm and captured the general public's imagination. From a technical standpoint, it also marked a drastic change in the favored architecture to design generative image models. GANs used to be the de facto choice, with techniques like StyleGAN. With DALL.E 2, autoregressive and diffusion models became the new standard for large-scale generative models overnight. This rapid shift raises a fundamental question: can we scale up GANs to benefit from large datasets like LAION? We find that naïvely increasing the capacity of the StyleGan architecture quickly becomes unstable. We introduce GigaGAN, a new GAN architecture that far exceeds this limit, demonstrating GANs as a viable option for text-to-image synthesis. GigaGAN offers three major advantages. First, it is orders of magnitude faster at inference time, taking only 0.13 seconds to synthesize a 512px image. Second, it can synthesize high-resolution images, for example, 16-megapixel images in 3.66 seconds. Finally, GigaGAN supports various latent space editing applications such as latent interpolation, style mixing, and vector arithmetic operations.

Neural networks have shown great abilities in estimating depth from a single image. However, the inferred depth maps are well below one-megapixel resolution and often lack fine-grained details, which limits their practicality. Our method builds on our analysis on how the input resolution and the scene structure affects depth estimation performance. We demonstrate that there is a trade-off between a consistent scene structure and the high-frequency details, and merge low- and high-resolution estimations to take advantage of this duality using a simple depth merging network. We present a double estimation method that improves the whole-image depth estimation and a patch selection method that adds local details to the final result. We demonstrate that by merging estimations at different resolutions with changing context, we can generate multi-megapixel depth maps with a high level of detail using a pre-trained model.

Abstract Image‐Based Rendering allows users to easily capture a scene using a single camera and then navigate freely with realistic results. However, the resulting renderings are completely static, and dynamic effects – such as fire, waterfalls or small waves – cannot be reproduced. We tackle the challenging problem of enabling free‐viewpoint navigation including such stationary dynamic effects, but still maintaining the simplicity of casual capture. Using a single camera – instead of previous complex synchronized multi‐camera setups – means that we have unsynchronized videos of the dynamic effect from multiple views, making it hard to blend them when synthesizing novel views. We present a solution that allows smooth free‐viewpoint video‐based rendering (VBR) of such scenes using temporal Laplacian pyramid decomposition video, enabling spatio‐temporal blending. For effects such as fire and waterfalls, that are semi‐transparent and occupy 3D space, we first estimate their spatial volume. This allows us to create per‐video geometries and alpha‐matte videos that we can blend using our frequency‐dependent method. We also extend Laplacian blending to the temporal dimension to remove additional temporal seams. We show results on scenes containing fire, waterfalls or rippling waves at the seaside, bringing these scenes to life.

This paper describes a simple technique to analyze Generative Adversarial Networks (GANs) and create interpretable controls for image synthesis, such as change of viewpoint, aging, lighting, and time of day. We identify important latent directions based on Principal Component Analysis (PCA) applied either in latent space or feature space. Then, we show that a large number of interpretable controls can be defined by layer-wise perturbation along the principal directions. Moreover, we show that BigGAN can be controlled with layer-wise inputs in a StyleGAN-like manner. We show results on different GANs trained on various datasets, and demonstrate good qualitative matches to edit directions found through earlier supervised approaches.

Many digital design tasks require a user to set a large number of parameters. Gallery-based interfaces provide a way to quickly evaluate examples and explore the space of potential designs, but require systems to predict which designs from a high-dimensional space are the right ones to present to the user. In this paper we present the design adjectives framework for building parameterized design tools in high dimensional design spaces. The framework allows users to create and edit design adjectives, machine-learned models of user intent, to guide exploration through high-dimensional design spaces. We provide a domain-agnostic implementation of the design adjectives framework based on Gaussian process regression, which is able to rapidly learn user intent from only a few examples. Learning and sampling of the design adjective occurs at interactive rates, making the system suitable for iterative design workflows. We demonstrate use of the design adjectives framework to create design tools for three domains: materials, fonts, and particle systems. We evaluate these tools in a user study showing that participants were able to easily explore the design space and find designs that they liked, and in professional case studies that demonstrate the framework's ability to support professional design concepting workflows.

Abstract Lighting is a critical element of theater. A lighting designer is responsible for drawing the audience's attention to a specific part of the stage, setting time of day, creating a mood, and conveying emotions. Designers often begin the lighting design process by collecting reference visual imagery that captures different aspects of their artistic intent. Then, they experiment with various lighting options to determine which ideas work best on stage. However, modern stages contain tens to hundreds of lights, and setting each light source's parameters individually to realize an idea is both tedious and requires expert skill. In this paper, we describe an exploratory lighting design tool based on feedback from professional designers. The system extracts abstract visual objectives from reference imagery and applies them to target regions of the stage. Our system can rapidly generate plausible design candidates that embody the visual objectives through a Gibbs sampling method, and present them as a design gallery for rapid exploration and iterative refinement. We demonstrate that the resulting system allows lighting designers of all skill levels to quickly create and communicate complex designs, even for scenes containing many color‐changing lights.

Abstract Copying an element from a photo and pasting it into a painting is a challenging task. Applying photo compositing techniques in this context yields subpar results that look like a collage — and existing painterly stylization algorithms, which are global, perform poorly when applied locally. We address these issues with a dedicated algorithm that carefully determines the local statistics to be transferred. We ensure both spatial and inter‐scale statistical consistency and demonstrate that both aspects are key to generating quality results. To cope with the diversity of abstraction levels and types of paintings, we introduce a technique to adjust the parameters of the transfer depending on the painting. We show that our algorithm produces significantly better results than photo compositing or global stylization techniques and that it enables creative painterly edits that would be otherwise difficult to achieve.

Accurate representation of soft transitions between image regions is essential for high-quality image editing and compositing. Current techniques for generating such representations depend heavily on interaction by a skilled visual artist, as creating such accurate object selections is a tedious task. In this work, we introduce semantic soft segments , a set of layers that correspond to semantically meaningful regions in an image with accurate soft transitions between different objects. We approach this problem from a spectral segmentation angle and propose a graph structure that embeds texture and color features from the image as well as higher-level semantic information generated by a neural network. The soft segments are generated via eigendecomposition of the carefully constructed Laplacian matrix fully automatically. We demonstrate that otherwise complex image editing tasks can be done with little effort using semantic soft segments.

Lighting is a critical element of portrait photography. However, good lighting design typically requires complex equipment and significant time and expertise. Our work simplifies this task using a relighting technique that transfers the desired illumination of one portrait onto another. The novelty in our approach to this challenging problem is our formulation of relighting as a mass transport problem. We start from standard color histogram matching that only captures the overall tone of the illumination, and we show how to use the mass-transport formulation to make it dependent on facial geometry. We fit a three-dimensional (3D) morphable face model to the portrait, and for each pixel, we combine the color value with the corresponding 3D position and normal. We then solve a mass-transport problem in this augmented space to generate a color remapping that achieves localized, geometry-aware relighting. Our technique is robust to variations in facial appearance and small errors in face reconstruction. As we demonstrate, this allows our technique to handle a variety of portraits and illumination conditions, including scenarios that are challenging for previous methods.

This paper introduces a deep-learning approach to photographic style transfer that handles a large variety of image content while faithfully transferring the reference style. Our approach builds upon the recent work on painterly transfer that separates style from the content of an image by considering different layers of a neural network. However, as is, this approach is not suitable for photorealistic style transfer. Even when both the input and reference images are photographs, the output still exhibits distortions reminiscent of a painting. Our contribution is to constrain the transformation from the input to the output to be locally affine in colorspace, and to express this constraint as a custom fully differentiable energy term. We show that this approach successfully suppresses distortion and yields satisfying photorealistic style transfers in a broad variety of scenarios, including transfer of the time of day, weather, season, and artistic edits.

Intrinsic images are a mid-level representation of an image that decompose the image into reflectance and illumination layers. The reflectance layer captures the color/texture of surfaces in the scene, while the illumination layer captures shading effects caused by interactions between scene illumination and surface geometry. Intrinsic images have a long history in computer vision and recently in computer graphics, and have been shown to be a useful representation for tasks ranging from scene understanding and reconstruction to image editing. In this report, we review and evaluate past work on this problem. Specifically, we discuss each work in terms of the priors they impose on the intrinsic image problem. We introduce a new synthetic ground-truth dataset that we use to evaluate the validity of these priors and the performance of the methods. Finally, we evaluate the performance of the different methods in the context of image-editing applications.

Visual formats have advanced beyond single-view images and videos: 3D movies are commonplace, researchers have developed multi-view navigation systems, and VR is helping to push light field cameras to mass market. However, editing tools for these media are still nascent, and even simple filtering operations like color correction or stylization are problematic: naively applying image filters per frame or per view rarely produces satisfying results due to time and space inconsistencies. Our method preserves and stabilizes filter effects while being agnostic to the inner working of the filter. It captures filter effects in the gradient domain, then uses input frame gradients as a reference to impose temporal and spatial consistency. Our least-squares formulation adds minimal overhead compared to naive data processing. Further, when filter cost is high, we introduce a filter transfer strategy that reduces the number of per-frame filtering computations by an order of magnitude, with only a small reduction in visual quality. We demonstrate our algorithm on several camera array formats including stereo videos, light fields, and wide baselines.

Lighting is a critical element of portrait photography. However, good lighting design typically requires complex equipment and significant time and expertise. Our work simplifies this task using a relighting technique that transfers the desired illumination of one portrait onto another. The novelty in our approach to this challenging problem is our formulation of relighting as a mass transport problem. We start from standard color histogram matching that only captures the overall tone of the illumination, and we show how to use the mass-transport formulation to make it dependent on facial geometry. We fit a three-dimensional (3D) morphable face model to the portrait, and for each pixel, we combine the color value with the corresponding 3D position and normal. We then solve a mass-transport problem in this augmented space to generate a color remapping that achieves localized, geometry-aware relighting. Our technique is robust to variations in facial appearance and small errors in face reconstruction. As we demonstrate, this allows our technique to handle a variety of portraits and illumination conditions, including scenarios that are challenging for previous methods.

We propose a method to remove objects such as people and cars from multi-view urban image datasets, enabling free-viewpoint IBR in the edited scenes. Our method combines information from multi-view 3D reconstruction with image inpainting techniques, by formulating the problem as an optimization of a global patch-based objective function. We use Image-Based Rendering (IBR) techniques to reproject information from neighboring views, and 3D multi-view stereo reconstruction to perform multiview coherent initialization for inpainting of pixels not filled by reprojection. Our algorithm performs multi-view consistent inpainting for color and 3D by blending reprojections with patch-based image inpainting. We run our algorithm on casually captured datasets, and Google StreetViewdata, removing objects cars, people and pillars, showing that our approach produces results of sufficient quality for free-viewpoint IBR on "cleaned up" scenes, as well as IBR scene editing, such as limited motion of real objects.

The growth of online marketplaces for selling goods has increased the need for product photography by novice users and consumers. Additionally, the increased use of online media and large-screen billboards promotes the adoption of videos for advertising, going beyond just using still imagery. Lighting is a key distinction between professional and casual product videography. Professionals use specialized hardware setups, and bring expert skills to create good lighting that shows off the product's shape and material, while also producing aesthetically pleasing results. In this paper, we introduce a new do-it-yourself (DIY) approach to lighting design that lets novice users create studio quality product videography. We identify design principles to light products through emphasizing highlights, rim lighting, and contours. We devise a set of computational metrics to achieve these design goals. Our workflow is: the user acquires a video of the product by mounting a video camera on a tripod and using a tablet to light objects by waving the tablet around the object. We automatically analyze and split this acquired video into snippets that match our design principles. Finally, we present an interface that lets users easily select snippets with specific characteristics and then assembles them to produce a final pleasing video of the product. Alternatively, they can rely on our template mechanism to automatically assemble a video.

Abstract Portraiture is a major art form in both photography and painting. In most instances, artists seek to make the subject stand out from its surrounding, for instance, by making it brighter or sharper. In the digital world, similar effects can be achieved by processing a portrait image with photographic or painterly filters that adapt to the semantics of the image. While many successful user‐guided methods exist to delineate the subject, fully automatic techniques are lacking and yield unsatisfactory results. Our paper first addresses this problem by introducing a new automatic segmentation algorithm dedicated to portraits. We then build upon this result and describe several portrait filters that exploit our automatic segmentation algorithm to generate high‐quality portraits.

Demosaicking and denoising are the key first stages of the digital imaging pipeline but they are also a severely ill-posed problem that infers three color values per pixel from a single noisy measurement. Earlier methods rely on hand-crafted filters or priors and still exhibit disturbing visual artifacts in hard cases such as moiré or thin edges. We introduce a new data-driven approach for these challenges: we train a deep neural network on a large corpus of images instead of using hand-tuned filters. While deep learning has shown great success, its naive application using existing training datasets does not give satisfactory results for our problem because these datasets lack hard cases. To create a better training set, we present metrics to identify difficult patches and techniques for mining community photographs for such patches. Our experiments show that this network and training procedure outperform state-of-the-art both on noisy and noise-free data. Furthermore, our algorithm is an order of magnitude faster than the previous best performing techniques.

Photo retouching enables photographers to invoke dramatic visual impressions by artistically enhancing their photos through stylistic color and tone adjustments. However, it is also a time-consuming and challenging task that requires advanced skills beyond the abilities of casual photographers. Using an automated algorithm is an appealing alternative to manual work, but such an algorithm faces many hurdles. Many photographic styles rely on subtle adjustments that depend on the image content and even its semantics. Further, these adjustments are often spatially varying. Existing automatic algorithms are still limited and cover only a subset of these challenges. Recently, deep learning has shown unique abilities to address hard problems. This motivated us to explore the use of deep neural networks (DNNs) in the context of photo editing. In this article, we formulate automatic photo adjustment in a manner suitable for this approach. We also introduce an image descriptor accounting for the local semantics of an image. Our experiments demonstrate that training DNNs using these descriptors successfully capture sophisticated photographic styles. In particular and unlike previous techniques, it can model local adjustments that depend on image semantics. We show that this yields results that are qualitatively and quantitatively better than previous work.

There is often more than one way to select tonal adjustment-for a photograph, and different individuals may prefer different adjustments. However, selecting good adjustments is challenging. This paper describes a method to predict whether a given tonal rendition is acceptable for a photograph, which we use to characterize its range of acceptable adjustments. We gathered a dataset of image “acceptability” over brightness and contrast adjustments. We find that unacceptable renditions can be explained in terms of overexposure, under-exposure, and low contrast. Based on this observation, we propose a machine-learning algorithm to assess whether an adjusted photograph looks acceptable. We show that our algorithm can differentiate unsightly renditions from reasonable ones. Finally, we describe proof-of-concept applications that use our algorithm to guide the exploration of the possible tonal renditions of a photograph.

In this talk, I present two projects that are inspired by how photographers work. These projects are in collaboration with Ivo Boyadzhiev and Kavita Bala at Cornell University and Ted Adelson at MIT.

Highly optimized programs are prone to bit rot, where performance quickly becomes suboptimal in the face of new hardware and compiler techniques. In this paper we show how to automatically lift performance-critical stencil kernels from a stripped x86 binary and generate the corresponding code in the high-level domain-specific language Halide. Using Halide’s state-of-the-art optimizations targeting current hardware, we show that new optimized versions of these kernels can replace the originals to rejuvenate the application for newer hardware. The original optimized code for kernels in stripped binaries is nearly impossible to analyze statically. Instead, we rely on dynamic traces to regenerate the kernels. We perform buffer structure reconstruction to identify input, intermediate and output buffer shapes. We abstract from a forest of concrete dependency trees which contain absolute memory addresses to symbolic trees suitable for high-level code generation. This is done by canonicalizing trees, clustering them based on structure, inferring higher-dimensional buffer accesses and finally by solving a set of linear equations based on buffer accesses to lift them up to simple, high-level expressions. Helium can handle highly optimized, complex stencil kernels with input-dependent conditionals. We lift seven kernels from Adobe Photoshop giving a 75% performance improvement, four kernels from IrfanView, leading to 4.97× performance, and one stencil from the miniGMG multigrid benchmark netting a 4.25× improvement in performance. We manually rejuvenated Photoshop by replacing eleven of Photoshop’s filters with our lifted implementations, giving 1.12× speedup without affecting the user experience.

Abstract In the digital world, assigning arbitrary colors to an object is a simple operation thanks to texture mapping. However, in the real world, the same basic function of applying colors onto an object is far from trivial. One can specify colors during the fabrication process using a color 3D printer, but this does not apply to already existing objects. Paint and decals can be used during post‐fabrication, but they are challenging to apply on complex shapes. In this paper, we develop a method to enable texture mapping of physical objects, that is, we allow one to map an arbitrary color image onto a three‐dimensional object. Our approach builds upon hydrographics, a technique to transfer pigments printed on a sheet of polymer onto curved surfaces. We first describe a setup that makes the traditional water transfer printing process more accurate and consistent across prints. We then simulate the transfer process using a specialized parameterization to estimate the mapping between the planar color map and the object surface. We demonstrate that our approach enables the application of detailed color maps onto complex shapes such as 3D models of faces and anatomical casts.

Extending image processing techniques to videos is a non-trivial task; applying processing independently to each video frame often leads to temporal inconsistencies, and explicitly encoding temporal consistency requires algorithmic changes. We describe a more general approach to temporal consistency. We propose a gradient-domain technique that is blind to the particular image processing algorithm. Our technique takes a series of processed frames that suffers from flickering and generates a temporally-consistent video sequence. The core of our solution is to infer the temporal regularity from the original unprocessed video, and use it as a temporal consistency guide to stabilize the processed sequence. We formally characterize the frequency properties of our technique, and demonstrate, in practice, its ability to stabilize a wide range of popular image processing techniques including enhancement and stylization of color and tone, intrinsic images, and depth estimation.

Photographers often “prep” their subjects to achieve various effects; for example, toning down overly shiny skin, covering blotches, etc. Making such adjustments digitally after a shoot is possible, but difficult without good tools and good skills. Making such adjustments to video footage is harder still. We describe and study a set of 2D image operations, based on multiscale image analysis, that are easy and straightforward and that can consistently modify perceived material properties. These operators first build a subband decomposition of the image and then selectively modify the coefficients within the subbands. We call this selection process band sifting . We show that different siftings of the coefficients can be used to modify the appearance of properties such as gloss, smoothness, pigmentation, or weathering. The band-sifting operators have particularly striking effects when applied to faces; they can provide “knobs” to make a face look wetter or drier, younger or older, and with heavy or light variation in pigmentation. Through user studies, we identify a set of operators that yield consistent subjective effects for a variety of materials and scenes. We demonstrate that these operators are also useful for processing video sequences.

Editing materials in photos opens up numerous opportunities like turning an unappealing dirt ground into luscious grass and creating a comfortable wool sweater in place of a cheap t-shirt. However, such edits are challenging. Approaches such as 3D rendering and BTF rendering can represent virtually everything, but they are also data intensive and computationally expensive, which makes user interaction difficult. Leaner methods such as texture synthesis are more easily controllable by artists, but also more limited in the range of materials that they handle, for example, grass and wool are typically problematic because of their non-Lambertian reflectance and numerous self-occlusions. We propose a new approach for editing of complex materials in photographs. We extend the texture-by-numbers approach with ideas from texture interpolation. The inputs to our method are coarse user annotation maps that specify the desired output, such as the local scale of the material and the illumination direction. Our algorithm then synthesizes the output from a discrete set of annotated exemplars. A key component of our method is that it can cope with missing data, interpolating information from the available exemplars when needed. This enables production of satisfying results involving materials with complex appearance variations such as foliage, carpet, and fabric from only one or a couple of exemplar photographs.

Cloud image processing is often proposed as a solution to the limited computing power and battery life of mobile devices: it allows complex algorithms to run on powerful servers with virtually unlimited energy supply. Unfortunately, this overlooks the time and energy cost of uploading the input and downloading the output images. When transfer overhead is accounted for, processing images on a remote server becomes less attractive and many applications do not benefit from cloud offloading. We aim to change this in the case of image enhancements that preserve the overall content of an image. Our key insight is that, in this case, the server can compute and transmit a description of the transformation from input to output, which we call a transform recipe. At equivalent quality, our recipes are much more compact than JPEG images: this reduces the client's download. Furthermore, recipes can be computed from highly compressed inputs which significantly reduces the data uploaded to the server. The client reconstructs a high-fidelity approximation of the output by applying the recipe to its local high-quality input. We demonstrate our results on 168 images and 10 image processing applications, showing that our recipes form a compact representation for a diverse set of image filters. With an equivalent transmission budget, they provide higher-quality results than JPEG-compressed input/output images, with a gain of the order of 10 dB in many cases. We demonstrate the utility of recipes on a mobile phone by profiling the energy consumption and latency for both local and cloud computation: a transform recipe-based pipeline runs 2--4x faster and uses 2--7x less energy than local or naive cloud computation.

Multiscale manipulations are central to image editing but also prone to halos. Achieving artifact-free results requires sophisticated edge-aware techniques and careful parameter tuning. These shortcomings were recently addressed by the local Laplacian filters, which can achieve a broad range of effects using standard Laplacian pyramids. However, these filters are slow to evaluate and their relationship to other approaches is unclear. In this article, we show that they are closely related to anisotropic diffusion and to bilateral filtering. Our study also leads to a variant of the bilateral filter that produces cleaner edges while retaining its speed. Building upon this result, we describe an acceleration scheme for local Laplacian filters on gray-scale images that yields speedups on the order of 50×. Finally, we demonstrate how to use local Laplacian filters to alter the distribution of gradients in an image. We illustrate this property with a robust algorithm for photographic style transfer.

Headshot portraits are a popular subject in photography but to achieve a compelling visual style requires advanced skills that a casual photographer will not have. Further, algorithms that automate or assist the stylization of generic photographs do not perform well on headshots due to the feature-specific, local retouching that a professional photographer typically applies to generate such portraits. We introduce a technique to transfer the style of an example headshot photo onto a new one. This can allow one to easily reproduce the look of renowned artists. At the core of our approach is a new multiscale technique to robustly transfer the local statistics of an example portrait onto a new one. This technique matches properties such as the local contrast and the overall lighting direction while being tolerant to the unavoidable differences between the faces of two different people. Additionally, because artists sometimes produce entire headshot collections in a common style, we show how to automatically find a good example to use as a reference for a given portrait, enabling style transfer without the user having to search for a suitable example for each input. We demonstrate our approach on data taken in a controlled environment as well as on a large set of photos downloaded from the Internet. We show that we can successfully handle styles by a variety of different artists.

State-of-the-art single image deblurring techniques are sensitive to image noise. Even a small amount of noise, which is inevitable in low-light conditions, can degrade the quality of blur kernel estimation dramatically. The recent approach of Tai and Lin [17] tries to iteratively denoise and deblur a blurry and noisy image. However, as we show in this work, directly applying image denoising methods often partially damages the blur information that is extracted from the input image, leading to biased kernel estimation. We propose a new method for handling noise in blind image deconvolution based on new theoretical and practical insights. Our key observation is that applying a directional low-pass filter to the input image greatly reduces the noise level, while preserving the blur information in the orthogonal direction to the filter. Based on this observation, our method applies a series of directional filters at different orientations to the input image, and estimates an accurate Radon transform of the blur kernel from each filtered image. Finally, we reconstruct the blur kernel using inverse Radon transform. Experimental results on synthetic and real data show that our algorithm achieves higher quality results than previous approaches on blurry and noisy images.

Image processing pipelines combine the challenges of stencil computations and stream programs. They are composed of large graphs of different stencil stages, as well as complex reductions, and stages with global or data-dependent access patterns. Because of their complex structure, the performance difference between a naive implementation of a pipeline and an optimized one is often an order of magnitude. Efficient implementations require optimization of both parallelism and locality, but due to the nature of stencils, there is a fundamental tension between parallelism, locality, and introducing redundant recomputation of shared values.

Abstract Selections are central to image editing, e.g., they are the starting point of common operations such as copy‐pasting and local edits. Creating them by hand is particularly tedious and scribble‐based techniques have been introduced to assist the process. By interpolating a few strokes specified by users, these methods generate precise selections. However, most of the algorithms assume a100% accurate input, and even small inaccuracies in the scribbles often degrade the selection quality, which imposes an additional burden on users. In this paper, we propose a selection technique tolerant to input inaccuracies. We use a dense conditional random field (CRF) to robustly infer a selection from possibly inaccurate input. Further, we show that patch‐based pixel similarity functions yield more precise selection than simple point‐wise metrics. However, efficiently solving a dense CRF is only possible in low‐dimensional Euclidean spaces, and the metrics that we use are high‐dimensional and often non‐Euclidean. We address this challenge by embedding pixels in a low‐dimensional Euclidean space with a metric that approximates the desired similarity function. The results show that our approach performs better than previous techniques and that two options are sufficient to cover a variety of images depending on whether the objects are textured.

In most professional cinema productions, the color palette of the movie is painstakingly adjusted by a team of skilled colorists -- through a process referred to as color grading -- to achieve a certain visual look. The time and expertise required to grade a video makes it difficult for amateurs to manipulate the colors of their own video clips. In this work, we present a method that allows a user to transfer the color palette of a model video clip to their own video sequence. We estimate a per-frame color transform that maps the color distributions in the input video sequence to that of the model video clip. Applying this transformation naively leads to artifacts such as bleeding and flickering. Instead, we propose a novel differential-geometry-based scheme that interpolates these transformations in a manner that minimizes their curvature, similarly to curvature flows. In addition, we automatically determine a set of keyframes that best represent this interpolated transformation curve, and can be used subsequently, to manually refine the color grade. We show how our method can successfully transfer color palettes between videos for a range of visual styles and a number of input video clips.

Good lighting is crucial in photography and can make the difference between a great picture and a discarded image. Traditionally, professional photographers work in a studio with many light sources carefully set up, with the goal of getting a near-final image at exposure time, with post-processing mostly focusing on aspects orthogonal to lighting. Recently, a new workflow has emerged for architectural and commercial photography, where photographers capture several photos from a fixed viewpoint with a moving light source. The objective is not to produce the final result immediately, but rather to capture useful data that are later processed, often significantly, in photo editing software to create the final well-lit image. This new workflow is flexible, requires less manual setup, and works well for time-constrained shots. But dealing with several tens of unorganized layers is painstaking, requiring hours to days of manual effort, as well as advanced photo editing skills. Our objective in this paper is to make the compositing step easier. We describe a set of optimizations to assemble the input images to create a fewbasis lightsthat correspond to common goals pursued by photographers, e.g., accentuating edges and curved regions. We also introducemodifiersthat capture standard photographic tasks, e.g., to alter the lights to soften highlights and shadows, akin to umbrellas and soft boxes. Our experiments with novice and professional users show that our approach allows them to quickly create satisfying results, whereas working with unorganized images requires considerably more time. Casual users particularly benefit from our approach since coping with a large number of layers is daunting for them and requires significant experience.

We introduce "time hallucination": synthesizing a plausible image at a different time of day from an input image. This challenging task often requires dramatically altering the color appearance of the picture. In this paper, we introduce the first data-driven approach to automatically creating a plausible-looking photo that appears as though it were taken at a different time of day. The time of day is specified by a semantic time label, such as "night". Our approach relies on a database of time-lapse videos of various scenes. These videos provide rich information about the variations in color appearance of a scene throughout the day. Our method transfers the color appearance from videos with a similar scene as the input photo. We propose a locally affine model learned from the video for the transfer, allowing our model to synthesize new color data while retaining image details. We show that this model can hallucinate a wide range of different times of day. The model generates a large sparse linear system, which can be solved by off-the-shelf solvers. We validate our methods by synthesizing transforming photos of various outdoor scenes to four times of interest: daytime, the golden hour, the blue hour, and nighttime.

Reconstructing realistic 3D hair geometry is challenging due to omnipresent occlusions, complex discontinuities and specular appearance. To address these challenges, we propose a multi-view hair reconstruction algorithm based on orientation fields with structure-aware aggregation. Our key insight is that while hair's color appearance is view-dependent, the response to oriented filters that captures the local hair orientation is more stable. We apply the structure-aware aggregation to the MRF matching energy to enforce the structural continuities implied from the local hair orientations. Multiple depth maps from the MRF optimization are then fused into a globally consistent hair geometry with a template refinement procedure. Compared to the state-of-the-art color-based methods, our method faithfully reconstructs detailed hair structures. We demonstrate the results for a number of hair styles, ranging from straight to curly, and show that our framework is suitable for capturing hair in motion.

Abstract Optical flow is a critical component of video editing applications, e.g. for tasks such as object tracking, segmentation, and selection. In this paper, we propose an optical flow algorithm calledSimpleFlowwhose running times increase sublinearly in the number of pixels. Central to our approach is a probabilistic representation of the motion flow that is computed using only local evidence and without resorting to global optimization. To estimate the flow in image regions where the motion is smooth, we use a sparse set of samples only, thereby avoiding the expensive computation inherent in traditional dense algorithms. We show that our results can be used as is for a variety of video editing tasks. For applications where accuracy is paramount, we use our result to bootstrap a global optimization. This significantly reduces the running times of such methods without sacrificing accuracy. We also demonstrate that the SimpleFlow algorithm can process HD and 4K footage in reasonable times.

Proper white balance is essential in photographs to eliminate color casts due to illumination. The single-light case is hard to solve automatically but relatively easy for humans. Unfortunately, many scenes contain multiple light sources such as an indoor scene with a window, or when a flash is used in a tungsten-lit room. The light color can then vary on a per-pixel basis and the problem becomes challenging at best, even with advanced image editing tools. We propose a solution to the ill-posed mixed light white balance problem, based on user guidance. Users scribble on a few regions that should have the same color, indicate one or more regions of neutral color, and select regions where the current color looks correct. We first expand the provided scribble groups to more regions using pixel similarity and a robust voting scheme. We formulate the spatially varying white balance problem as a sparse data interpolation problem in which the user scribbles and their extensions form constraints. We demonstrate that our approach can produce satisfying results on a variety of scenes with intuitive scribbles and without any knowledge about the lights.

An intrinsic image is a decomposition of a photo into an illumination layer and a reflectance layer, which enables powerful editing such as the alteration of an object's material independently of its illumination. However, decomposing a single photo is highly under-constrained and existing methods require user assistance or handle only simple scenes. In this paper, we compute intrinsic decompositions using several images of the same scene under different viewpoints and lighting conditions. We use multi-view stereo to automatically reconstruct 3D points and normals from which we derive relationships between reflectance values at different locations, across multiple views and consequently different lighting conditions. We use robust estimation to reliably identify reflectance ratios between pairs of points. From these, we infer constraints for our optimization and enforce a coherent solution across multiple views and illuminations. Our results demonstrate that this constrained optimization yields high-quality and coherent intrinsic decompositions of complex scenes. We illustrate how these decompositions can be used for image-based illumination transfer and transitions between views with consistent lighting.

Using existing programming tools, writing high-performance image processing code requires sacrificing readability, portability, and modularity. We argue that this is a consequence of conflating what computations define the algorithm , with decisions about storage and the order of computation. We refer to these latter two concerns as the schedule , including choices of tiling, fusion, recomputation vs. storage, vectorization, and parallelism. We propose a representation for feed-forward imaging pipelines that separates the algorithm from its schedule, enabling high-performance without sacrificing code clarity. This decoupling simplifies the algorithm specification: images and intermediate buffers become functions over an infinite integer domain, with no explicit storage or boundary conditions. Imaging pipelines are compositions of functions. Programmers separately specify scheduling strategies for the various functions composing the algorithm, which allows them to efficiently explore different optimizations without changing the algorithmic code. We demonstrate the power of this representation by expressing a range of recent image processing applications in an embedded domain specific language called Halide, and compiling them for ARM, x86, and GPUs. Our compiler targets SIMD units, multiple cores, and complex memory hierarchies. We demonstrate that it can handle algorithms such as a camera raw pipeline, the bilateral grid, fast local Laplacian filtering, and image segmentation. The algorithms expressed in our language are both shorter and faster than state-of-the-art implementations.

Adjusting photographs to obtain compelling renditions requires skill and time. Even contrast and brightness adjustments are challenging because they require taking into account the image content. Photographers are also known for having different retouching preferences. As the result of this complexity, rule-based, one-size-fits-all automatic techniques often fail. This problem can greatly benefit from supervised machine learning but the lack of training data has impeded work in this area. Our first contribution is the creation of a high-quality reference dataset. We collected 5,000 photos, manually annotated them, and hired 5 trained photographers to retouch each picture. The result is a collection of 5 sets of 5,000 example input-output pairs that enable supervised learning. We first use this dataset to predict a user's adjustment from a large training set. We then show that our dataset and features enable the accurate adjustment personalization using a carefully chosen set of training photos. Finally, we introduce difference learning: this method models and predicts difference between users. It frees the user from using predetermined photos for training. We show that difference learning enables accurate prediction using only a handful of examples.

Adjusting photographs to obtain compelling renditions requires skill and time. Even contrast and brightness adjustments are challenging because they require taking into account the image content. Photographers are also known for having different retouching preferences. As the result of this complexity, rule-based, one-size-fits-all automatic techniques often fail. This problem can greatly benefit from supervised machine learning but the lack of training data has impeded work in this area. Our first contribution is the creation of a high-quality reference dataset. We collected 5,000 photos, manually annotated them, and hired 5 trained photographers to retouch each picture. The result is a collection of 5 sets of 5,000 example input-output pairs that enable supervised learning. We first use this dataset to predict a user's adjustment from a large training set. We then show that our dataset and features enable the accurate adjustment personalization using a carefully chosen set of training photos. Finally, we introduce difference learning: this method models and predicts difference between users. It frees the user from using predetermined photos for training. We show that difference learning enables accurate prediction using only a handful of examples.

Camera shake is a common source of degradation in photographs. Restoring blurred pictures is challenging because both the blur kernel and the sharp image are unknown, which makes this problem severely underconstrained. In this work, we estimate camera shake by analyzing edges in the image, effectively constructing the Radon transform of the kernel. Building upon this result, we describe two algorithms for estimating spatially invariant blur kernels. In the first method, we directly invert the transform, which is computationally efficient since it is not necessary to also estimate the latent sharp image. This approach is well suited for scenes with a diversity of edges, such as man-made environments. In the second method, we incorporate the Radon transform within the MAP estimation framework to jointly estimate the kernel and the image. While more expensive, this algorithm performs well on a broader variety of scenes, even when fewer edges can be observed. Our experiments show that our algorithms achieve comparable results to the state of the art in general and produce superior outputs on man-made scenes and photos degraded by a small kernel.

This paper introduces the video mesh, a data structure for representing video as 2.5D “paper cutouts.” The video mesh allows interactive editing of moving objects and modeling of depth, which enables 3D effects and post-exposure camera control. The video mesh sparsely encodes optical flow as well as depth, and handles occlusion using local layering and alpha mattes. Motion is described by a sparse set of points tracked over time. Each point also stores a depth value. The video mesh is a triangulation over this point set and per-pixel information is obtained by interpolation. The user rotoscopes occluding contours and we introduce an algorithm to cut the video mesh along them. Object boundaries are refined with per-pixel alpha values. The video mesh is at its core a set of texture mapped triangles, we leverage graphics hardware to enable interactive editing and rendering of a variety of effects. We demonstrate the effectiveness of our representation with special effects such as 3D viewpoint changes, object insertion, depth-of-field manipulation, and 2D to 3D video conversion.

We describe a set of image editing and viewing tools that explicitly take into account the resolution of the display on which the image is viewed. Our approach is twofold. First, we design editing tools that process only the visible data, which is useful for images larger than the display. This encompasses cases such as multi-image panoramas and high-resolution medical data. Second, we propose an adaptive way to set viewing parameters such brightness and contrast. Because we deal with very large images, different locations and scales often require different viewing parameters. We let users set these parameters at a few places and interpolate satisfying values everywhere else. We demonstrate the efficiency of our approach on different display and image sizes. Since the computational complexity to render a view depends on the display resolution and not the actual input image resolution, we achieve interactive image editing even on a 16 gigapixel image.

Photo deblurring has been a major research topic in the past few years. So far, existing methods have focused on removing the blur due to camera shake and object motion. In this paper, we show that the optical system of the camera also generates significant blur, even with professional lenses. We introduce a method to estimate the blur kernel densely over the image and across multiple aperture and zoom settings. Our measures show that the blur kernel can have a non-negligible spread, even with top-of-the-line equipment, and that it varies nontrivially over this domain. In particular, the spatial variations are not radially symmetric and not even left-right symmetric. We develop and compare two models of the optical blur, each of them having its own advantages. We show that our models predict accurate blur kernels that can be used to restore photos. We demonstrate that we can produce images that are more uniformly sharp unlike those produced with spatially-invariant deblurring techniques.

Computer Aided Design (CAD) typically involves tasks such as adjusting the camera perspective and assembling pieces in free space that require specifying 6 degrees of freedom (DOF). The standard approach is to factor these DOFs into 2D subspaces that are mapped to the x and y axes of a mouse. This metaphor is inherently modal because one needs to switch between subspaces, and disconnects the input space from the modeling space. In this paper, we propose a bimanual hand tracking system that provides physically-motivated 6-DOF control for 3D assembly. First, we discuss a set of principles that guide the design of our precise, easy-to-use, and comfortable-to-use system. Based on these guidelines, we describe a 3D input metaphor that supports constraint specification classically used in CAD software, is based on only a few simple gestures, lets users rest their elbows on their desk, and works alongside the keyboard and mouse. Our approach uses two consumer-grade webcams to observe the user's hands. We solve the pose estimation problem with efficient queries of a precomputed database that relates hand silhouettes to their 3D configuration. We demonstrate efficient 3D mechanical assembly of several CAD models using our hand-tracking system.

Interpolation between pairs of values, typically vectors, is a fundamental operation in many computer graphics applications. In some cases simple linear interpolation yields meaningful results without requiring domain knowledge. However, interpolation between pairs of distributions or pairs of functions often demands more care because features may exhibit translational motion between exemplars. This property is not captured by linear interpolation. This paper develops the use of displacement interpolation for this class of problem, which provides a generic method for interpolating between distributions or functions based on advection instead of blending. The functions can be non-uniformly sampled, high-dimensional, and defined on non-Euclidean manifolds, e.g., spheres and tori. Our method decomposes distributions or functions into sums of radial basis functions (RBFs). We solve a mass transport problem to pair the RBFs and apply partial transport to obtain the interpolated function. We describe practical methods for computing the RBF decomposition and solving the transport problem. We demonstrate the interpolation approach on synthetic examples, BRDFs, color distributions, environment maps, stipple patterns, and value functions.

The Laplacian pyramid is ubiquitous for decomposing images into multiple scales and is widely used for image analysis. However, because it is constructed with spatially invariant Gaussian kernels, the Laplacian pyramid is widely believed as being unable to represent edges well and as being ill-suited for edge-aware operations such as edge-preserving smoothing and tone mapping. To tackle these tasks, a wealth of alternative techniques and representations have been proposed, e.g., anisotropic diffusion, neighborhood filtering, and specialized wavelet bases. While these methods have demonstrated successful results, they come at the price of additional complexity, often accompanied by higher computational cost or the need to post-process the generated results. In this paper, we show state-of-the-art edge-aware processing using standard Laplacian pyramids. We characterize edges with a simple threshold on pixel values that allows us to differentiate large-scale edges from small-scale details. Building upon this result, we propose a set of image filters to achieve edge-preserving smoothing, detail enhancement, tone mapping, and inverse tone mapping. The advantage of our approach is its simplicity and flexibility, relying only on simple point-wise nonlinearities and small Gaussian convolutions; no optimization or post-processing is required. As we demonstrate, our method produces consistently high-quality results, without degrading edges or introducing halos.

Abstract Significant progress has been made in high‐quality hair rendering, but it remains difficult to choose parameter values that reproduce a given real hair appearance. In particular, for applications such as games where naive users want to create their own avatars, tuning complex parameters is not practical. Our approach analyses a single flash photograph and estimates model parameters that reproduce the visual likeness of the observed hair. The estimated parameters include color absorptions, three reflectance lobe parameters of a multiple‐scattering rendering model, and a geometric noise parameter. We use a novel melanin‐based model to capture the natural subspace of hair absorption parameters. At its core, the method assumes that images of hair with similar color distributions are also similar in appearance. This allows us to recast the issue as an image retrieval problem where the photo is matched with a dataset of rendered images; we thus also match the model parameters used to generate these images. An earth‐mover's distance is used between luminance‐weighted color distributions to gauge similarity. We conduct a perceptual experiment to evaluate this metric in the context of hair appearance and demonstrate the method on 64 photographs, showing that it can achieve a visual likeness for a large variety of input photos.

For many computational photography applications, the lighting and materials in the scene are critical pieces of information. We seek to obtainintrinsic images, which decompose a photo into the product of anilluminationcomponent that represents lighting effects and areflectancecomponent that is the color of the observed material. This is an under-constrained problem and automatic methods are challenged by complex natural images. We describe a new approach that enables users to guide an optimization with simple indications such as regions of constant reflectance or illumination. Based on a simple assumption on local reflectance distributions, we derive a new propagation energy that enables a closed form solution using linear least-squares. We achieve fast performance by introducing a novel downsampling that preserves local color distributions. We demonstrate intrinsic image decomposition on a variety of images and show applications.

Artists often need to import and embellish 3D models coming from CAD-CAM into 2D vector graphics software to produce, e.g., brochures or manuals. Current automatic solutions tend to result, at best, in a 2D triangle soup and artists often have to trace over 3D renderings. We describe a method to convert 3D models into 2D layered vector illustrations that respect visibility and facilitate further editing. Our core contribution is a visibility method that can partition a mesh into large components that can be layered according to visibility. Because self-occluding objects and objects forming occlusion cycles cannot be represented by layers without being cut, we introduce a new cut algorithm that uses a graph representation of the mesh and curvature-aware geodesic distances.

White balance is a crucial step in the photographic pipeline. It ensures the proper rendition of images by eliminating color casts due to differing illuminants. Digital cameras and editing programs provide white balance tools that assume a single type of light per image, such as daylight. However, many photos are taken under mixed lighting. We propose a white balance technique for scenes with two light types that are specified by the user. This covers many typical situations involving indoor/outdoor or flash/ambient light mixtures. Since we work from a single image, the problem is highly underconstrained. Our method recovers a set of dominant material colors which allows us to estimate the local intensity mixture of the two light types. Using this mixture, we can neutralize the light colors and render visually pleasing images. Our method can also be used to achieve post-exposure relighting effects.

We accurately capture the shape and appearance of a person's hairstyle. We use triangulation and a sweep with planes of light for the geometry. Multiple projectors and cameras address the challenges raised by the reflectance and intricate geometry of hair. We introduce the use of structure tensors to infer the hidden geometry between the hair surface and the scalp. Our triangulation approach affords substantial accuracy improvement and we are able to measure elaborate hair geometry including complex curls and concavities. To reproduce the hair appearance, we capture a six-dimensional reflectance field. We introduce a new reflectance interpolation technique that leverages an analytical reflectance model to alleviate cross-fading artifacts caused by linear methods. Our results closely match the real hairstyles and can be used for animation.

Mean shift is a popular method to segment images and videos. Pixels are represented by feature points, and the segmentation is driven by the point density in feature space. In this paper, we introduce the use of Morse theory to interpret mean shift as a topological decomposition of the feature space into density modes. This allows us to build on the watershed technique and design a new algorithm to compute mean-shift segmentations of images and videos. In addition, we introduce the use of topological persistence to create a segmentation hierarchy. We validated our method by clustering images using color cues. In this context, our technique runs faster than previous work, especially on videos and large images. We evaluated accuracy with a classical benchmark which shows results on par with existing low-level techniques, i.e. we do not sacrifice accuracy for speed.

We introduce a new surface representation method, called patchwork, to extend three-dimensional surface reconstruction capabilities from multiple images. A patchwork is the combination of several patches that are built one by one. This design potentially allows for the reconstruction of an object with arbitrarily large dimensions while preserving a fine level of detail. We formally demonstrate that this strategy leads to a spatial complexity independent of the dimensions of the reconstructed object and to a time complexity that is linear with respect to the object area. The former property ensures that we never run out of storage and the latter means that reconstructing an object can be done in a reasonable amount of time. In addition, we show that the patchwork representation handles equivalently open and closed surfaces, whereas most of the existing approaches are limited to a specific scenario, an open or closed surface, but not both. The patchwork concept is orthogonal to the method chosen for surface optimization. Most of the existing optimization techniques can be cast into this framework. To illustrate the possibilities offered by this approach, we propose two applications that demonstrate how our method dramatically extends a recent accurate graph technique based on minimal cuts. We first revisit the popular carving techniques. This results in a well-posed reconstruction problem that still enjoys the tractability of voxel space. We also show how we can advantageously combine several image-driven criteria to achieve a finely detailed geometry by surface propagation. These two examples demonstrate the versatility and flexibility of patchwork reconstruction. They underscore other properties inherited from patchwork representation: Although some min-cut methods have difficulty in handling complex shapes (e.g., with complex topologies), they can naturally manipulate any geometry through the patchwork representation while preserving their intrinsic qualities. The above properties of patchwork representation and reconstruction are demonstrated with real image sequences.

We present a new data structure---the bilateral grid , that enables fast edge-aware image processing. By working in the bilateral grid, algorithms such as bilateral filtering, edge-aware painting, and local histogram equalization become simple manipulations that are both local and independent. We parallelize our algorithms on modern GPUs to achieve real-time frame rates on high-definition video. We demonstrate our method on a variety of applications such as image editing, transfer of photographic look, and contrast enhancement of medical images.

We introduce a new approach to tone management for photographs. Whereas traditional tone-mapping operators target a neutral and faithful rendition of the input image, we explore pictorial looks by controlling visual qualities such as the tonal balance and the amount of detail. Our method is based on a two-scale non-linear decomposition of an image. We modify the different layers based on their histograms and introduce a technique that controls the spatial variation of detail. We introduce a Poisson correction that prevents potential gradient reversal and preserves detail. In addition to directly controlling the parameters, the user can transfer the look of a model photograph to the picture being edited.

This paper introduces a new method for surface reconstruction from multiple calibrated images. The primary contribution of this work is the notion of local prior to combine the flexibility of the carving approach with the accuracy of graph-cut optimization. A progressive refinement scheme is used to recover the topology and reason the visibility of the object. Within each voxel, a detailed surface patch is optimally reconstructed using a graph-cut method. The advantage of this technique is its ability to handle complex shape similarly to level sets while enjoying a higher precision. Compared to carving techniques, the addressed problem is well-posed, and the produced surface does not suffer from aliasing. In addition, our approach seamlessly handles complete and partial reconstructions: If the scene is only partially visible, the process naturally produces an open surface; otherwise, if the scene is fully visible, it creates a complete shape. These properties are demonstrated on real image sequences

Hair is a major feature of digital characters. Unfortunately, it has a complex geometry which challenges standard modeling tools. Some dedicated techniques exist, but creating a realistic hairstyle still takes hours. Complementary to user-driven methods, we here propose an image-based approach to capture the geometry of hair.The novelty of this work is that we draw information from the scattering properties of the hair that are normally considered a hindrance. To do so, we analyze image sequences from a fixed camera with a moving light source. We first introduce a novel method to compute the image orientation of the hairs from their anisotropic behavior. This method is proven to subsume and extend existing work while improving accuracy. This image orientation is then raised into a 3D orientation by analyzing the light reflected by the hair fibers. This part relies on minimal assumptions that have been proven correct in previous work.Finally, we show how to use several such image sequences to reconstruct the complete hair geometry of a real person. Results are shown to illustrate the fidelity of the captured geometry to the original hair. This technique paves the way for a new approach to digital hair generation.

We present a method to relight human faces in real time, using consumer-grade graphics cards even with limited 3D capabilities. We show how to render faces using a combination of a simple, hardware-accelerated parametric model simulating skin shading and a detail texture map, and provide robust procedures to estimate all the necessary parameters for a given face. Our model strikes a balance between the difficulty of realistic face rendering (given the very specific reflectance properties of skin) and the goal of real-time rendering with limited hardware capabilities. This is accomplished by automatically generating an optimal set of parameters for a simple rendering model. We offer a discussion of the issues in face rendering to discern the pros and cons of various rendering models and to generalize our approach to most of the current hardware constraints. We provide results demonstrating the usability of our approach and the improvements we introduce both in the performance and in the visual quality of the resulting faces.

This paper addresses the problem of 3D reconstruction from a set of viewpoints on a short baseline. Its main contribution is the development of a robust algorithm which can extract 3D informations from a set of images taken with a very small baseline, even in the presence of significant occlusion. To achieve this goal, we use a multi-pass process that works layer by layer. In each pass we begin with a "space carving" step which is made robust through some morphological operations. Then we introduce an original technique to solve the ambiguity inherent to "space carving" in the case of short baseline. Once we have a voxel representation of the objects, we propose a method based on differential geometry to build a smooth mesh. Results are presented, demonstrating the efficiency and usefulness of the method.