USPTO Patent US12586270B2: Image Editing with Generative Neural Network

ChangeBridge: Patent Grants - AI & Computing (G06N)

Published March 24th, 2026

Detected March 25th, 2026

Summary

The USPTO has granted patent US12586270B2 to Adobe Inc. for a method of generating and modifying digital images using a generative neural network. The patent covers systems and methods for latent-based editing of digital images by mapping feature tensors and style vectors into a joint feature style space.

View original document View source feed page

What changed

The United States Patent and Trademark Office (USPTO) has granted patent US12586270B2 to Adobe Inc. The patent, titled 'Image editing using generative neural network,' details systems, computer-readable media, and methods for latent-based editing of digital images. Specifically, it describes a process where feature tensors and style vectors are mapped into a joint feature style space, allowing for modifications to generate altered digital images using a generative neural network.

This patent grant represents a new intellectual property asset for Adobe Inc. in the field of AI-driven image manipulation. For technology companies operating in AI and digital imaging, this patent may influence their research and development strategies, particularly concerning generative models and image editing techniques. Compliance officers should note this as a development in the IP landscape related to AI technologies, though it does not impose direct regulatory obligations on entities outside of potential patent infringement considerations.

Source document (simplified)

← USPTO Patent Grants

Generating and modifying digital images using a joint feature style latent space of a generative neural network

Grant US12586270B2 Kind: B2 Mar 24, 2026

Assignee

Adobe Inc.

Inventors

Hui Qu, Baldo Faieta, Cameron Smith, Elya Shechtman, Jingwan Lu, Ratheesh Kalarot, Richard Zhang, Saeid Motiian, Shabnam Ghadar, Wei-An Lin

Abstract

The present disclosure relates to systems, non-transitory computer-readable media, and methods for latent-based editing of digital images using a generative neural network. In particular, in one or more embodiments, the disclosed systems perform latent-based editing of a digital image by mapping a feature tensor and a set of style vectors for the digital image into a joint feature style space. In one or more implementations, the disclosed systems apply a joint feature style perturbation and/or modification vectors within the joint feature style space to determine modified style vectors and a modified feature tensor. Moreover, in one or more embodiments the disclosed systems generate a modified digital image utilizing a generative neural network from the modified style vectors and the modified feature tensor.