US 12,136,169 B1
Generating a view of an object from existing images
Seyedpooya Mirhosseini, Port Jefferson, NY (US); and Seyedkoosha Mirhosseini, Cupertino, CA (US)
Assigned to APPLE INC., Cupertino, CA (US)
Filed by Apple Inc., Cupertino, CA (US)
Filed on Jan. 24, 2022, as Appl. No. 17/582,223.
Claims priority of provisional application 63/143,126, filed on Jan. 29, 2021.
Int. Cl. G06T 17/20 (2006.01); G06T 1/60 (2006.01); G06T 3/18 (2024.01); G06T 7/55 (2017.01); G06T 15/00 (2011.01)
CPC G06T 17/20 (2013.01) [G06T 1/60 (2013.01); G06T 3/18 (2024.01); G06T 7/55 (2017.01); G06T 15/005 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
at a device including, a non-transitory memory and one or more processors coupled with the non-transitory memory:
obtaining a request to view an object from a target point-of-view (POV), wherein the object is represented in a plurality of images captured from corresponding POVs that are different from the target POV;
generating respective contribution scores for the corresponding POVs, wherein each contribution score is a function of an amount of overlap between a view frustum of the corresponding POV and a view frustum of the target POV;
determining a sequence in which the plurality of images is ordered based on the respective contribution scores for the corresponding POVs; and
synthesizing a new view of the object corresponding to the target POV by:
performing a warping operation to the plurality of images in accordance with the sequence in order to generate sets of re-projection values; and
selecting pixel values for the target POV from the sets of re-projection values by inputting the sets of re-projection values into a decision function and receiving the pixel values for the target POV as an output of the decision function.