Apparatuses, systems, and techniques are presented to perform monocular view synthesis of a dynamic scene. Single and multi-view depth information can be determined for a collection of images of a dynamic scene, and a blender network can be used to combine image features for foreground, background, and missing image regions using fused depth maps inferred form the single and multi-view depth information.

Apparatuses, systems, and techniques are presented to perform monocular view synthesis of a dynamic scene. Single and multi-view depth information can be determined for a collection of images of a dynamic scene, and a blender network can be used to combine image features for foreground, background, and missing image regions using fused depth maps inferred form the single and multi-view depth information.

An exemplary system is configured to detect user input directing a telestration element to be drawn within an image depicting a surface within a scene; render, based on depth data representative of a depth map for the scene and within a three dimensional (3D) image depicting the surface within the scene, the telestration element; record a 3D position within the scene at which the telestration element is rendered within the 3D image; detect a telestration termination event that removes the telestration element from being rendered within the 3D image; and indicate, subsequent to the telestration termination event, an option to again render the telestration element at the 3D position.

The present disclosure describes systems and methods for improving binocular vision, which generate a virtual image moving between two different depths to stimulate and then strengthen the weaker/abnormal eye of the viewer to eventually improve or even restore his/her binocular vision based on the viewer's eye information. The system comprises an eye tracking module and a virtual image module. The eye tracking module is configured to provide eye information of the viewer. The virtual image module configured to display a first virtual object by projecting multiple normal light signals to a viewer's first eye to form a normal image and corresponding multiple adjusted light signals to a viewers second eye to form an adjusted image.

One disclosed example provides a method for displaying a hologram via a head-mounted display (HMD) device. The method comprises, via a camera system on the HMD device, acquiring image data capturing a surrounding environment by detecting illumination light output by a projector located on a case for the HMD device. A distance is determined from the HMD device to an object in the surrounding environment based upon the image data. The method further comprises displaying via the HMD device a hologram, the hologram comprising a left-eye image and a right-eye image each having a perspective based upon the distance determined.

Augmented and virtual reality display systems increase viewer comfort by reducing viewer exposure to virtual content that causes undesirable accommodation-vergence mismatches (AVM). The display systems limit displaying content that exceeds an accommodation-vergence mismatch threshold, which may define a volume around the viewer. The volume may be subdivided into two or more zones, including an innermost loss-of-fusion zone (LoF) in which no content is displayed, and one or more outer AVM zones in which the displaying of content may be stopped, or clipped, under certain conditions. For example, content may be clipped if the viewer is verging within an AVM zone and if the content is displayed within the AVM zone for more than a threshold duration. A further possible condition for clipping content is that the user is verging on that content. In addition, the boundaries of the AVM zone and/or the acceptable amount of time that the content is displayed may vary depending upon the type of content being displayed, e.g., whether the content is user-locked content or in-world content.

In one aspect, the provided is an authentication control device including: receiving means for receiving an input associated with at least one position among a plurality of positions that are included in an authentication image causing a viewer to perceive depth and are at different apparent depths; and determination means for determining, based on the input, whether or not the input is made by a human.

In one aspect, the provided is an authentication control device including: receiving means for receiving an input associated with at least one position among a plurality of positions that are included in an authentication image causing a viewer to perceive depth and are at different apparent depths; and determination means for determining, based on the input, whether or not the input is made by a human.

This disclosure relates to techniques for generating robust depth estimations for captured images using semantic segmentation. Semantic segmentation may be defined as a process of creating a mask over an image, wherein pixels are segmented into a predefined set of semantic classes. Such segmentations may be binary (e.g., a ‘person pixel’ or a ‘non-person pixel’) or multi-class (e.g., a pixel may be labelled as: ‘person,’ ‘dog,’ ‘cat,’ etc.). As semantic segmentation techniques grow in accuracy and adoption, it is becoming increasingly important to develop methods of utilizing such segmentations and developing flexible techniques for integrating segmentation information into existing computer vision applications, such as depth and/or disparity estimation, to yield improved results in a wide range of image capture scenarios. In some embodiments, an optimization framework may be employed to optimize a camera device's initial scene depth/disparity estimates that employs both semantic segmentation and color regularization in a robust fashion.

This disclosure relates to techniques for generating robust depth estimations for captured images using semantic segmentation. Semantic segmentation may be defined as a process of creating a mask over an image, wherein pixels are segmented into a predefined set of semantic classes. Such segmentations may be binary (e.g., a ‘person pixel’ or a ‘non-person pixel’) or multi-class (e.g., a pixel may be labelled as: ‘person,’ ‘dog,’ ‘cat,’ etc.). As semantic segmentation techniques grow in accuracy and adoption, it is becoming increasingly important to develop methods of utilizing such segmentations and developing flexible techniques for integrating segmentation information into existing computer vision applications, such as depth and/or disparity estimation, to yield improved results in a wide range of image capture scenarios. In some embodiments, an optimization framework may be employed to optimize a camera device's initial scene depth/disparity estimates that employs both semantic segmentation and color regularization in a robust fashion.