US 12,307,596 B2
3D garment generation from 2D scribble images
Panagiotis Achlioptas, Mountain View, CA (US); Menglei Chai, Los Angeles, CA (US); Hsin-Ying Lee, San Jose, CA (US); Kyle Olszewski, Los Angeles, CA (US); Jian Ren, Hermosa Beach, CA (US); and Sergey Tulyakov, Santa Monica, CA (US)
Assigned to Snap Inc., Santa Monica, CA (US)
Filed by Panagiotis Achlioptas, Mountain View, CA (US); Menglei Chai, Los Angeles, CA (US); Hsin-Ying Lee, San Jose, CA (US); Kyle Olszewski, Los Angeles, CA (US); Jian Ren, Hermosa Beach, CA (US); and Sergey Tulyakov, Santa Monica, CA (US)
Filed on Sep. 30, 2022, as Appl. No. 17/957,049.
Prior Publication US 2024/0112401 A1, Apr. 4, 2024
Int. Cl. G06T 19/00 (2011.01); G06T 17/10 (2006.01)
CPC G06T 17/10 (2013.01) [G06T 19/00 (2013.01); G06T 2210/16 (2013.01); G06T 2210/56 (2013.01)] 16 Claims
OG exemplary drawing
 
1. A method of generating a three-dimensional (3D) garment image from a two-dimensional (2D) garment image drawn by a user, the method comprising:
receiving the 2D garment image drawn by the user;
translating, by a conditional 2D generator, the 2D garment image into color-density pairs comprising a red, green, blue (RGB) projection and a density map of a local density at each pixel generated from front and rear views of the 2D garment image;
conducting, by a density-aware sampling algorithm, density-aware sampling of point locations and colors of corresponding pairs of the color-density pairs to generate a flat point cloud;
translating, by a conditional 3D generator, the flat point cloud into an output 3D point cloud; and
sending the output 3D point cloud for display as the 3D garment image.