Abstract: Methods, system, and computer storage media are provided for generating physical-based materials for rendering digital objects with an appearance of a real-world material. Images depicted the real-world material, including diffuse component images and specular component images, are captured using different lighting patterns, which may include area lights. From the captured images, approximations of one or more material maps are determined using a photometric stereo technique. Based on the approximations and the captured images, a neural network system generates a set of material maps, such as a diffuse albedo material map, a normal material map, a specular albedo material map, and a roughness material map. The material maps from the neural network may be optimized based on a comparison of the input images of the real-world material and images rendered from the material maps.

Abstract: Embodiments are disclosed for neural texture mapping. In some embodiments, a method of neural texture mapping includes obtaining a plurality of images of an object, determining volumetric representation of a scene of the object using a first neural network, mapping 3D points of the scene to a 2D texture space using a second neural network, and determining radiance values for each 2D point in the 2D texture space from a plurality of viewpoints using a second neural network to generate a 3D appearance representation of the object.

Abstract: Techniques and systems are provided for configuring neural networks to perform warping of an object represented in an image to create a caricature of the object. For instance, in response to obtaining an image of an object, a warped image generator generates a warping field using the image as input. The warping field is generated using a model trained with pairings of training images and known warped images using supervised learning techniques and one or more losses. The warped image generator determines, based on the warping field, a set of displacements associated with pixels of the input image. These displacements indicate pixel displacement directions for the pixels of the input image. These displacements are applied to the digital image to generate a warped image of the object.

Abstract: Methods and systems are provided for facilitating large-scale augmented reality in relation to outdoor scenes using estimated camera pose information. In particular, camera pose information for an image can be estimated by matching the image to a rendered ground-truth terrain model with known camera pose information. To match images with such renders, data driven cross-domain feature embedding can be learned using a neural network. Cross-domain feature descriptors can be used for efficient and accurate feature matching between the image and the terrain model renders. This feature matching allows images to be localized in relation to the terrain model, which has known camera pose information. This known camera pose information can then be used to estimate camera pose information in relation to the image.

Abstract: Images can be edited to include features similar to a different target image. An unconditional generative adversarial network (GAN) is employed to edit features of an initial image based on a constraint determined from a target image. The constraint used by the GAN is determined from keypoints or segmentation masks of the target image, and edits are made to features of the initial image based on keypoints or segmentation masks of the initial image corresponding to those of the constraint from the target image. The GAN modifies the initial image based on a loss function having a variable for the constraint. The result of this optimization process is a modified initial image having features similar to the target image subject to the constraint determined from the identified keypoints or segmentation masks.

Abstract: Techniques and systems are provided for configuring neural networks to perform warping of an object represented in an image to create a caricature of the object. For instance, in response to obtaining an image of an object, a warped image generator generates a warping field using the image as input. The warping field is generated using a model trained with pairings of training images and known warped images using supervised learning techniques and one or more losses. The warped image generator determines, based on the warping field, a set of displacements associated with pixels of the input image. These displacements indicate pixel displacement directions for the pixels of the input image. These displacements are applied to the digital image to generate a warped image of the object.

Abstract: Images can be edited to include features similar to a different target image. An unconditional generative adversarial network (GAN) is employed to edit features of an initial image based on a constraint determined from a target image. The constraint used by the GAN is determined from keypoints or segmentation masks of the target image, and edits are made to features of the initial image based on keypoints or segmentation masks of the initial image corresponding to those of the constraint from the target image. The GAN modifies the initial image based on a loss function having a variable for the constraint. The result of this optimization process is a modified initial image having features similar to the target image subject to the constraint determined from the identified keypoints or segmentation masks.

Abstract: Embodiments of the present invention provide systems, methods, and non-transitory computer storage media for generating an ambient occlusion (AO) map for a 2D image that can be combined with the 2D image to adjust the contrast of the 2D image based on the geometric information in the 2D image. In embodiments, using a trained neural network, an AO map for a 2D image is automatically generated without any predefined 3D scene information. Optimizing the neural network to generate an estimated AO map for a 2D image requires training, testing, and validating the neural network using a synthetic dataset comprised of pairs of images and ground truth AO maps rendered from 3D scenes. By using an estimated AO map to adjust the contrast of a 2D image, the contrast of the image can be adjusted to make the image appear lifelike by modifying the shadows and shading in the image based on the ambient lighting present in the image.

Abstract: The present disclosure is directed toward systems and methods for predicting lighting conditions. In particular, the systems and methods described herein analyze a single low-dynamic range digital image to estimate a set of high-dynamic range lighting conditions associated with the single low-dynamic range lighting digital image. Additionally, the systems and methods described herein train a convolutional neural network to extrapolate lighting conditions from a digital image. The systems and methods also augment low-dynamic range information from the single low-dynamic range digital image by using a sky model algorithm to predict high-dynamic range lighting conditions.

Abstract: Embodiments of the present invention provide systems, methods, and computer storage media directed to generating training image data for a convolutional neural network, encoding parameters into a convolutional neural network, and employing a convolutional neural network that estimates camera calibration parameters of a camera responsible for capturing a given digital image. A plurality of different digital images can be extracted from a single panoramic image given a range of camera calibration parameters that correspond to a determined range of plausible camera calibration parameters. With each digital image in the plurality of extracted different digital images having a corresponding set of known camera calibration parameters, the digital images can be provided to the convolutional neural network to establish high-confidence correlations between detectable characteristics of a digital image and its corresponding set of camera calibration parameters.

Abstract: Methods and systems are provided for determining high-dynamic range lighting parameters for input low-dynamic range images. A neural network system can be trained to estimate high-dynamic range lighting parameters for input low-dynamic range images. The high-dynamic range lighting parameters can be based on sky color, sky turbidity, sun color, sun shape, and sun position. Such input low-dynamic range images can be low-dynamic range panorama images or low-dynamic range standard images. Such a neural network system can apply the estimates high-dynamic range lighting parameters to objects added to the low-dynamic range images.

Abstract: Methods and systems are provided for determining high-dynamic range lighting parameters for input low-dynamic range images. A neural network system can be trained to estimate high-dynamic range lighting parameters for input low-dynamic range images. The high-dynamic range lighting parameters can be based on sky color, sky turbidity, sun color, sun shape, and sun position. Such input low-dynamic range images can be low-dynamic range panorama images or low-dynamic range standard images. Such a neural network system can apply the estimates high-dynamic range lighting parameters to objects added to the low-dynamic range images.

Abstract: This disclosure relates to methods, non-transitory computer readable media, and systems that can render a virtual object in a digital image by using a source-specific-lighting-estimation-neural network to generate three-dimensional (“3D”) lighting parameters specific to a light source illuminating the digital image. To generate such source-specific-lighting parameters, for instance, the disclosed systems utilize a compact source-specific-lighting-estimation-neural network comprising both common network layers and network layers specific to different lighting parameters. In some embodiments, the disclosed systems further train such a source-specific-lighting-estimation-neural network to accurately estimate spatially varying lighting in a digital image based on comparisons of predicted environment maps from a differentiable-projection layer with ground-truth-environment maps.

Abstract: The present disclosure is directed toward systems and methods for predicting lighting conditions. In particular, the systems and methods described herein analyze a single low-dynamic range digital image to estimate a set of high-dynamic range lighting conditions associated with the single low-dynamic range lighting digital image. Additionally, the systems and methods described herein train a convolutional neural network to extrapolate lighting conditions from a digital image. The systems and methods also augment low-dynamic range information from the single low-dynamic range digital image by using a sky model algorithm to predict high-dynamic range lighting conditions.

Abstract: The present disclosure is directed toward systems and methods for predicting lighting conditions. In particular, the systems and methods described herein analyze a single low-dynamic range digital image to estimate a set of high-dynamic range lighting conditions associated with the single low-dynamic range lighting digital image. Additionally, the systems and methods described herein train a convolutional neural network to extrapolate lighting conditions from a digital image. The systems and methods also augment low-dynamic range information from the single low-dynamic range digital image by using a sky model algorithm to predict high-dynamic range lighting conditions.

Abstract: Embodiments of the present invention provide systems, methods, and computer storage media directed to generating training image data for a convolutional neural network, encoding parameters into a convolutional neural network, and employing a convolutional neural network that estimates camera calibration parameters of a camera responsible for capturing a given digital image. A plurality of different digital images can be extracted from a single panoramic image given a range of camera calibration parameters that correspond to a determined range of plausible camera calibration parameters. With each digital image in the plurality of extracted different digital images having a corresponding set of known camera calibration parameters, the digital images can be provided to the convolutional neural network to establish high-confidence correlations between detectable characteristics of a digital image and its corresponding set of camera calibration parameters.

Abstract: Embodiments of the present invention provide systems, methods, and computer storage media directed to generating training image data for a convolutional neural network, encoding parameters into a convolutional neural network, and employing a convolutional neural network that estimates camera calibration parameters of a camera responsible for capturing a given digital image. A plurality of different digital images can be extracted from a single panoramic image given a range of camera calibration parameters that correspond to a determined range of plausible camera calibration parameters. With each digital image in the plurality of extracted different digital images having a corresponding set of known camera calibration parameters, the digital images can be provided to the convolutional neural network to establish high-confidence correlations between detectable characteristics of a digital image and its corresponding set of camera calibration parameters.

Abstract: Embodiments of the present invention provide systems, methods, and computer storage media directed to generating training image data for a convolutional neural network, encoding parameters into a convolutional neural network, and employing a convolutional neural network that estimates camera calibration parameters of a camera responsible for capturing a given digital image. A plurality of different digital images can be extracted from a single panoramic image given a range of camera calibration parameters that correspond to a determined range of plausible camera calibration parameters. With each digital image in the plurality of extracted different digital images having a corresponding set of known camera calibration parameters, the digital images can be provided to the convolutional neural network to establish high-confidence correlations between detectable characteristics of a digital image and its corresponding set of camera calibration parameters.

Abstract: The present disclosure is directed toward systems and methods for predicting lighting conditions. In particular, the systems and methods described herein analyze a single low-dynamic range digital image to estimate a set of high-dynamic range lighting conditions associated with the single low-dynamic range lighting digital image. Additionally, the systems and methods described herein train a convolutional neural network to extrapolate lighting conditions from a digital image. The systems and methods also augment low-dynamic range information from the single low-dynamic range digital image by using a sky model algorithm to predict high-dynamic range lighting conditions.