Expression Image Processing Model for Video Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing methods fail to enhance the interestingness of videos and images, which has become a pressing need with the advancement of video applications.
Innovation Solution
An image processing method that acquires an expression image, adjusts it based on a preset image processing model to generate a video with a change process of the expression, and displays the video, utilizing a trained image processing model that can modify smile degrees or eye opening/closing degrees, and a migration model that applies expression changes to facial regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a preset image processing model is used to adjust expressions in images, then the interestingness and user experience of videos are improved, but the device complexity and computational resources required increase
Solution Approach 1:
The image processing model is trained in advance offline to learn expression change patterns from multiple images. During actual video generation, the pre-trained model directly applies learned expression changes to new images, avoiding the need for complex real-time computation and reducing device complexity while maintaining high interestingness
Solution Approach 2:
The model learns from example images and their corresponding expression changes, creating a reusable template of expression dynamics. This copied knowledge is then applied to generate expression changes in target images, reducing the need for complex real-time analysis while preserving natural expression patterns
2Reliability
If multiple images are processed to generate consistent expression change trends, then the reliability and consistency of the video output are improved, but the processing time and productivity are reduced
Solution Approach 1:
The system pre-processes multiple reference images during the training phase to establish consistent expression change patterns. This preliminary analysis of multiple images creates a robust model that ensures reliability during actual video generation without requiring real-time processing of multiple images, thus maintaining productivity
Solution Approach 2:
The trained model continuously applies the same expression change pattern across different target images, ensuring consistent and reliable expression dynamics throughout the video. This continuous application of learned patterns maintains reliability while enabling efficient batch processing that preserves productivity
Data Source
AI summary
This disclosure relates to an image processing method, an image processing apparatus, a device, and a storage medium, wherein after acquiring an expression image, an expression in the expression image can be adjusted based on a preset image processing model to generate a video with a change process of the expression, and the video is displayed to the user.