US 11,538,096 B2 · Method, medium, and system for live preview via machine learning models

Abstract

Machine learning-based approaches are used to create instances or visualizations of content appearing within an object in an image. For example, a user may submit a request for a preview or visualization of content within an object or other media such as a glass crystal. A trained model can process the content to generate adjustment data or other data that can be used to control image blending operations. The adjustment data can be applied to the pixel values of the content to modify the content in order to enable a visualization of the content within an object. The image portion can be modified such that the object appears to “blend” with and appear within the object. Image transformation techniques can be used to project the modified content onto a representation of an object. Thereafter, a visualization or preview of the content within the representation of the object can be presented.

Claims20

Dependent claims

A computing system, comprising:

at least one computing processor;
and
memory including instructions that, when executed by the at least one computing processor, enable the computing system to:
receive a request to view an area of interest represented in a query image with an object;
obtain a trained image adjustment model configured to generate image adjustment data to modify pixel data;
evaluate the trained image adjustment model on the area of interest to generate image adjustment data for blending an object of interest from a foreground image with a target background region of a background image;
apply the image adjustment data to pixels values of the area of interest;
and
generate image data that includes a representation of what the area of interest looks like when the area of interest is one of etched or engraved into a surface of the object based on the blending, wherein the object is at least one of a crystal object or a glass object.

6 more elements

A non-transitory computer readable storage medium storing one or more sequences of instructions executable by one or more processors to perform a set of operations comprising:

receiving a request to view an area of interest represented in a query image with an object;
obtaining a trained image adjustment model configured to generate image adjustment data to modify pixel data;
evaluating the trained image adjustment model on the area of interest to generate image adjustment data for blending an object of interest from a foreground image with a target background region of a background image;
applying the image adjustment data to pixels values of the area of interest;
and
generating image data that includes a representation of what the area of interest looks like when the area of interest is one of etched or engraved into a surface of the object based on the blending, wherein the object is at least one of a crystal object or a glass object.

3 more elements

Dependencies are inferred from the claim text.

Method, medium, and system for live preview via machine learning models

Life of this patent

Method, medium, and system for live preview via machine learning models

Life of this patent

Abstract

Claims20