Abstract: Methods, non-transitory computer readable media, apparatuses, and systems for image inpainting include obtaining, via a user interface, an input image and a local content preservation value and receiving a content generation selection. The content generation selection applies the local content preservation value to at least one pixel of the input image. An image generation model then generates an output image based on the input image and the content generation selection. The output image includes synthetic content in a region specified by the content generation selection, and a degree of adherence of the synthetic content to the input image is based on the local content preservation value.

Abstract: Systems and methods for non-destructive image editing are described. Embodiments are configured to obtain, via a document editor user interface, a selection input identifying a portion of a first image displayed in the document editor user interface. According to some aspects, the first image is part of a first layer of a multilayer document. Embodiments are further configured to: obtain an image generation text prompt; generate, using an image generation network, a second image based on the first image and the image generation text prompt; and present, via a multilayer window of the document editor user interface, a first element representing the first layer of the multilayer document and a second element representing a second layer of the multilayer document. The first element includes the first image and the second element includes the second image and a mask corresponding to the portion of the first image.

Abstract: Systems and methods for image processing are provided. Embodiments of the present disclosure obtain an image and a target dimension for expanding the image. The system generates a prompt based on the image using a prompt generation network. A diffusion model generates an expanded image based on the image, the target dimension, and the prompt, where the expanded image includes additional content in an outpainted region that is consistent with content of the image and the prompt.

Abstract: One or more aspects of the method, apparatus, and non-transitory computer readable medium include obtaining an original image, a scene graph describing elements of the original image, and a description of a modification to the original image. The one or more aspects further include updating the scene graph based on the description of the modification. The one or more aspects further include generating a modified image using an image generation neural network based on the updated scene graph, wherein the modified image incorporates content based on the original image and the description of the modification.