US 12,086,901 B2
Generating digital paintings utilizing an intelligent painting pipeline for improved brushstroke sequences
Jaskirat Singh, Canberra (AU); Jose Ignacio Echevarria Vallespi, South San Francisco, CA (US); and Cameron Smith, Santa Cruz, CA (US)
Assigned to Adobe Inc., San Jose, CA (US)
Filed by Adobe Inc., San Jose, CA (US)
Filed on Mar. 29, 2022, as Appl. No. 17/656,907.
Prior Publication US 2023/0316590 A1, Oct. 5, 2023
Int. Cl. G06T 11/00 (2006.01); G06N 3/08 (2023.01); G06T 7/194 (2017.01); G06T 7/70 (2017.01)
CPC G06T 11/00 (2013.01) [G06N 3/08 (2013.01); G06T 7/194 (2017.01); G06T 7/70 (2017.01); G06T 2207/20021 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A non-transitory computer readable medium storing instructions that, when executed by at least one processor, cause a computing device to:
generate a painted digital image from a target digital image comprising one or more foreground objects and a background region utilizing an image painting model to perform progressive layering by:
generating, for a digital canvas of digital painting strokes, background strokes for a first painting layer in the progressive layering, wherein the background strokes portray a painted version of the background region of the target digital image;
generating foreground strokes for one or more successive painting layers added to the first painting layer in the digital canvas, wherein the foreground strokes portray painted versions of the one or more foreground objects in the target digital image;
comparing states of the digital canvas with the target digital image according to a layer-specific reward function that incorporates a foreground saliency map to indicate salient foreground objects; and
generating the painted digital image from the digital canvas by adding the foreground strokes of the one or more successive painting layers to the background strokes of the first painting layer based on comparing the states of the digital canvas with the target digital image using the layer-specific reward function.