Image Stylization With User Constraint Control for Higher Quality

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image stylization models can only generate images in one style and lack user control, resulting in low-quality styled images and poor user experience.

Innovation Solution

An image stylization processing method that involves obtaining an original image, displaying conditional constraint information, generating a target condition feature based on user operations on this information, and creating a target image using the feature.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If a single stylization model is used to convert images, then the model structure remains simple, but the image quality and user control capability deteriorate

Engineering Contradiction:
Improvemodel structureVSAvoidimage quality
Core Design Contradiction:
Device complexityVSManufacturing precision

Solution Approach 1:

The patent divides the single stylization model into multiple specialized models, each trained to generate images in a specific target style. This segmentation allows each model to specialize in one style, improving image quality and controllability for that style, while keeping each individual model relatively simple in structure.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent creates a multi-functional system where multiple stylization models work together to handle different target styles. The system achieves universality by being able to convert input images to various styles through selecting the appropriate specialized model, thus improving overall image quality and user control without requiring each model to be overly complex.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Ease of operation

If a single stylization model is used, then the system remains simple to operate, but user control capability and image quality deteriorate

Engineering Contradiction:
Improvesystem simplicityVSAvoiduser control capability
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The patent introduces dynamic model selection based on the desired target style. The system adapts by selecting the appropriate specialized stylization model according to the user's requirements, providing dynamic control capability. This allows users to control the output style reliably while the system manages the complexity of having multiple models behind the scenes.

Inventive Principle:
Principle #15Dynamics

3Manufacturing precision

If multiple stylization models are used to improve image quality, then the system complexity increases, but user control capability is needed to manage this complexity

Engineering Contradiction:
Improveimage qualityVSAvoidsystem complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent introduces a model selection mechanism that acts as an intermediary between the user and multiple stylization models. This intermediary manages the complexity by automatically selecting the appropriate model based on the target style requirement, allowing users to benefit from multiple specialized models without directly managing their complexity. The intermediary handles the coordination and selection, reducing the perceived system complexity for users.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP4723027A1Image stylization processing method and apparatus, device, storage medium and program product
Publication Date: 2026.04.08 BEIJING ZITIAO NETWORK TECH CO LTD
  • EP4723027A1 patent drawingFigure 1~2
  • EP4723027A1 patent drawingFigure 3a~3c
  • EP4723027A1 patent drawingFigure 3d~3f

AI summary

The present disclosure relates to a method, an apparatus, a device, a storage medium and a program product for image stylization processing. The method comprises: obtaining an original image input by a user; displaying at least one piece of conditional constraint information; in response to an operation on the at least one piece of conditional constraint information, generating a target condition feature based on the conditional constraint information corresponding to the operation; generating a target image based on the target condition feature.