Customized prop image generation method and apparatus

Patent metadata

Field	Value
Publication number	US-12573109-B2
Application number	US-202318400936-A
Country	US
Kind code	B2
Filing date	Dec 29, 2023
Priority date	Dec 30, 2022
Publication date	Mar 10, 2026
Grant date	Mar 10, 2026

Abstract

This application provides techniques for generating customized prop images. The techniques comprise obtaining an initial prop image that comprises a target object, wherein the initial prop image comprises an initial image of any prop among at least one prop held by a target online streamer; obtaining an action file configured to specify an action of the target object; parsing the action file to determine a target action corresponding to the action file; and generating a customized prop image by updating a current action of the target object in the initial prop image to the target action, wherein the customized prop image corresponds to the target online streamer.

First claim

What is claimed is:

1 . A method of generating customized prop images, comprising:
obtaining an initial prop image that comprises a target object, wherein the initial prop image comprises an initial image of any prop among at least one prop held by a target online streamer; obtaining an action file configured to specify an action of the target object, wherein the action file comprises at least one pose file or at least one expression file, and the action file further comprises at least one moving image generation file; parsing the action file to determine a target action corresponding to the action file, wherein the parsing the action file to determine a target action corresponding to the action file comprises:
generating at least one dynamic combination file based on the at least one moving image generation file and at least one of the at least one pose file or the at least one expression file, wherein the dynamic combination file comprises at least one of a pose file or an expression file, and the dynamic combination file comprises a moving image generation file, and
parsing the dynamic combination file to determine a target dynamic action corresponding to the dynamic combination file; and
generating a customized prop image by updating a current action of the target object in the initial prop image to the target action, wherein the generating a customized prop image by updating a current action of the target object in the initial prop image to the target action comprises generating a dynamic customized prop image by updating the current action of the target object in the initial prop image to the target dynamic action, and wherein the customized prop image corresponds to the target online streamer.

2 . The method according to claim 1 ,
wherein the action file comprises at least one pose file or at least one expression file, wherein the at least one pose file is configured to specify a limb action or a pose action of the target object, and wherein the at least one expression file is configured to specify a facial action or an expression action of the target object; wherein the parsing the action file to determine a target action corresponding to the action file comprises:
generating at least one static combination file based on the at least one pose file or the at least one expression file, and
parsing the at least one static combination file to determine a target static action corresponding to the at least one static combination file; and
wherein the generating a customized prop image by updating a current action of the target object in the initial prop image to the target action comprises: generating a static customized prop image by updating the current action of the target object in the initial prop image to the target static action.

3 . The method according to claim 2 , wherein the target static action comprises at least one of a target static pose or a target static expression, and wherein the generating a static customized prop image by updating the current action of the target object in the initial prop image to the target static action comprises:
generating the static customized prop image by replacing a current pose of the target object in the initial prop image with the target static pose; and generating the static customized prop image by replacing a current expression of the target object in the initial prop image with the target static expression.

4 . The method according to claim 1 ,
wherein the moving image generation file comprises information indicating a quantity of image frames in a sequence, and the quantity of image frames is greater than 1; and wherein the parsing the dynamic combination file to determine a target dynamic action corresponding to the dynamic combination file comprises:
obtaining an initial action corresponding to the target object in the initial image,
determining a reference action based on parsing the pose file or the expression file in the dynamic combination file, and
determining, based on the initial action and the reference action, the target dynamic action that comprises static actions in the quantity of image frames.

5 . The method according to claim 1 ,
wherein the moving image generation file comprises a plurality of pieces of sequence frame information, and each piece of sequence frame information indicates an action offset or an adjustment to a part of the target object in a corresponding image frame; and wherein the parsing the dynamic combination file to determine a target dynamic action corresponding to the dynamic combination file comprises:
determining a reference action based on parsing the pose file or the expression file in the dynamic combination file, and
generating a static action corresponding to each piece of sequence frame information by adjusting the reference action based on each piece of sequence frame information, wherein the target dynamic action comprises a plurality of static actions corresponding to the plurality of pieces of sequence frame information.

6 . The method according to claim 5 , wherein the generating a customized prop image by updating a current action of the target object in the initial prop image to the target action comprises:
generating a plurality of static images corresponding to the plurality of static actions comprised in the target dynamic action by replacing the current action of the target object in the initial prop image with each of the plurality of static actions; and generating a dynamic customized prop image by performing splicing processing on the plurality of static images.

7 . The method according to claim 1 , wherein after generating the customized prop image, the method further comprises:
identifying a theme color of the target object in the customized prop image; and setting a background picture of the target object in the customized prop image based on the theme color.

8 . The method according to claim 1 , wherein after generating the customized prop image, the method further comprises:
obtaining a digital collectible contract address; generating a request for creating a digital collectible corresponding to the customized prop image; and sending the request to a blockchain, wherein a digital collectible contract corresponding to the digital collectible contract address on the blockchain is configured to create the digital collectible corresponding to the customized prop image based on the request.

9 . The method according to claim 1 , wherein the initial prop image comprises a model resource required for constructing a model.

10 . A computing device, comprising a memory and a processor, wherein the memory stores computer-readable instructions that upon execution by the processor cause the processor to perform operations comprising:
obtaining an initial prop image that comprises a target object, wherein the initial prop image comprises an initial image of any prop among at least one prop held by a target online streamer; obtaining an action file configured to specify an action of the target object, wherein the action file comprises at least one pose file or at least one expression file, and the action file further comprises at least one moving image generation file; parsing the action file to determine a target action corresponding to the action file, wherein the parsing the action file to determine a target action corresponding to the action file comprises:
generating at least one dynamic combination file based on the at least one moving image generation file and at least one of the at least one pose file or the at least one expression file, wherein the dynamic combination file comprises at least one of a pose file or an expression file, and the

Assignees

• Shanghai Bilibili Tech Co Ltd

Inventors

• Na Zhicheng
• Chu Bo

Classifications

• G06T7/90

  Determination of colour characteristics · CPC title

• G06T2207/10024

  Color image · CPC title

• G06V40/20

  Movements or behaviour, e.g. gesture recognition (recognition of facial expressions G06V40/16) · CPC title

• G06V40/174 Primary

  Facial expression recognition · CPC title

• G06T2207/20221

  Image fusion; Image merging · CPC title