Abstract: In one embodiment, a method includes delineating a region in a two-dimensional (2D) image in a virtual reality environment, generating a mask based on an object detected in the delineated region, wherein the mask is defined by a perimeter, and generating a virtual object in the virtual reality environment, where the virtual object incorporates the detected object, and the virtual object is generated by applying the mask to a three-dimensional (3D) object. Delineating a region may include receiving user input that delineates the region, and the input may trace a perimeter of the region. Delineating a region may include detecting an object and determining a perimeter of the detected object. The detected object may be classified as corresponding to a specified object type. The 2D image may include a camera perspective view of an object in the virtual reality environment.
Type:
Grant
Filed:
December 30, 2016
Date of Patent:
May 12, 2020
Assignee:
Facebook, Inc.
Inventors:
Alexandros Alexander, Michael Stephen Booth
Abstract: Aspects of this disclosure are directed to updating a low resolution buffer during a binning pass of an image surface by a graphics processing unit (GPU). For example, during the binning pass of the image surface, the GPU may divide the image surface into a plurality of blocks of surface pixels of the image surface. The GPU may then store a surface identifier of the image surface within storage locations of the low resolution buffer to correspond to each of the blocks of the surface pixels of the image surface.