Granted Patent B2

US 12,254,561 · App. 17/706,420 · Granted Mar 18, 2025

Inventors: Evan Smyth (La Crescenta, CA); Gil Spencer (Incline Village, NV)

Assignee: SPREEAI CORPORATION

G06T15/205G06T7/70G06T13/40G06T2207/20084

Abstract

Claims (32)

1. A computer-implemented method for using an avatar to produce a digital image representation of a body of a user, said method comprising the steps of:

capturing, from one or more input images of the body from a video, measurements of the body and combining the measurements with an avatar to produce a set of avatar images;

invoking a neural network to generate a set of photorealistic face and hair images of the user;

combining the avatar images and the face and hair images to produce a set of composite images; and

converting the set of composite images into a set of final images representing the body by inputting, by the neural network, data describing an image sequence and matte representing a boundary between a head and a neck to facilitate accurate blending of image inputs.

2. The method of claim 1 wherein each image is a single still frame.

3. The method of claim 1 wherein the combining step further combines foreground images in addition to combining the avatar images and the face and hair images.

4. The method of claim 1 wherein the user is a human person.

5. The method of claim 4 wherein the digital image representation is a representation of the person's face, hair, body, garments, and shoes.

6. The method of claim 1 wherein the avatar images comprise at least one of garments and shoes.

7. The method of claim 1 wherein the step of invoking a neural network takes into account skin tone of the user, as well as illumination of the user's face and hair.

8. The method of claim 1 wherein the neural network inputs data describing an image sequence representing successive positions of an avatar head during playing of an avatar video.

9. The method of claim 1 wherein the neural network inputs data describing a sequence representing desired hair motion during playing of an avatar video.

10. The method of claim 1 wherein the neural network inputs data describing an image sequence representing illumination on a head area during playing of an avatar video.

11. The method of claim 1 wherein inputs to the neural network comprise a set of face and hair images.

12. The method of claim 1 wherein the neural network produces, in addition to face and hair images, an entire representation of the body, including the head of the body.

13. The method of claim 12 wherein the neural network produces measurements of the body.

14. Apparatus for producing a digital image representation of a body, said apparatus comprising one or more processors configured to:

capture, from one or more input images of the body from a video, measurements of the body and combine the measurements with an avatar to produce a set of avatar images;

generate, using a neural network, a set of photorealistic face and hair images;

combine the avatar images and the face and hair images to produce a set of composite images; and

convert the set of composite images into a set of final images representing the body by inputting, by the neural network, data describing an image sequence and matte representing a boundary between a head and a neck to facilitate accurate blending of image inputs.

15. The apparatus of claim 14 , wherein the one or more processors are further configured to convert a background scene into a set of background images and combine the set of background images along with the avatar images and the face and hair images.

16. The apparatus of claim 14 wherein the one or more processors are further configured to combine at least one of garments, shoes, accessories, background, jewelry, hats, eyeglasses, sunglasses, belts, purses, handbags, and backpacks associated with the body.

17. The apparatus of claim 14 wherein the one or more processors are further configured to add an audio track to the set of composite images.

18. One or more non-transitory computer-readable storage media storing instructions that, responsive to execution by a processing device, causes the processing device to:

capture, from one or more input images of the body from a video, measurements of the body and combine the measurements with an avatar to produce a set of avatar images;

invoke a neural network to generate a set of photorealistic face and hair images of the user;

combine the avatar images and the face and hair images to produce a set of composite images; and

convert the set of composite images into a set of final images representing the body by inputting, by the neural network, data describing an image sequence and matte representing a boundary between a head and a neck to facilitate accurate blending of image inputs.

19. The one or more non-transitory computer-readable storage media of claim 18 , wherein the instructions, responsive to execution by the processing device, further cause the processing device to convert a background scene into a set of background images and combine the set of background images along with the avatar images and the face and hair images.

20. The one or more non-transitory computer-readable storage media of claim 18 , wherein the instructions, responsive to execution by the processing device, further cause the processing device to combine at least one of garments, shoes, accessories, background, jewelry, hats, eyeglasses, sunglasses, belts, purses, handbags, and backpacks associated with the body.

Assignments (2)

Continuity (3)

References Cited (68)