Layered Image Editing for Multi-Object Rendering Overhead
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-object image editing methods incur significant performance overhead due to the need for re-rendering all objects when editing is revised, particularly in group photos where multiple human faces require different editing.
Innovation Solution
Render each editing object to separate layers, allowing independent editing operations on each layer without affecting others, thereby reducing the need for full-image re-rendering during revisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multi-object editing is performed on the original image by editing objects one by one, then editing functionality is achieved, but performance overhead increases enormously when re-editing objects
Solution Approach 1:
The original image is segmented into multiple independent layers, with each layer corresponding to a different editing object. This segmentation allows each object to be edited independently without affecting other objects, eliminating the need to re-render the entire image when re-editing a single object.
Solution Approach 2:
The patent transitions from editing objects within a single 2D image plane to editing objects across multiple stacked layers in a 3D-like structure. Each layer can be independently manipulated, and the final image is generated by compositing these layers, adding a dimensional aspect that enables independent object editing.
2Ease of operation
If all objects are rendered together on the original image, then editing results can be displayed, but re-editing requires re-rendering all objects increasing computational cost
Solution Approach 1:
The rendering process is segmented by creating separate layers for each editing object. Each layer is rendered independently and stored separately, allowing the system to display editing results without rendering all objects together. When re-editing is needed, only the specific layer requires re-rendering.
Solution Approach 2:
The patent performs preliminary rendering by creating separate layers for each editing object in advance. These pre-rendered layers are stored and can be quickly composed to display editing results without performing full re-rendering operations, reducing the time lost during re-editing operations.
3Quantity of substance
If intermediate images are not generated for each editing object, then storage space is saved, but re-editing requires re-rendering all other edited objects
Solution Approach 1:
Instead of storing a single intermediate image containing all edited objects, the patent segments the intermediate representation into multiple independent layers, with each layer corresponding to one editing object. This segmentation enables selective re-rendering of only the necessary layer during re-editing operations.
Solution Approach 2:
The patent creates layer copies for each editing object that can be independently manipulated and stored. These layer copies serve as intermediate representations that can be quickly re-rendered when needed, providing an efficient balance between storage requirements and re-editing performance.
Data Source
Figure 1~2
Figure 3~4(b)
Figure 5
AI summary
An image editing method and apparatus. The method comprises: acquiring an original image, wherein the original image comprises a first editing object and a second editing object, and the first editing object and the second editing object are located in different image areas of the original image; respectively rendering the first editing object and the second editing object to different image layers, so as to generate a first original image layer and a second original image layer; in response to a first editing operation for the first editing object, rendering an editing result, which is obtained by the first editing object under the first editing operation, on the basis of the first original image layer, so as to generate a first editing image layer; in response to a second editing operation for the second editing object, rendering an editing result, which is obtained by the second editing object under the second editing operation, on the basis of the second original image layer, so as to generate a second editing image layer; and generating a target image according to the first editing image layer and the second editing image layer.