Virtual Character Expression Driving With Left-Right Coefficient Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing expression driving methods result in inconsistency and asynchrony between related parts of a virtual character's face, leading to unrealistic driving effects and hindering user experience, particularly due to the lack of relational constraints on expression coefficients.
Innovation Solution
The method involves obtaining expression coefficients from a target image, dividing them into initial left and right coefficients, determining their similarity, and generating target coefficients based on this similarity to drive a virtual character, ensuring consistency and synchronization of facial expressions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If expression coefficients are obtained directly from image recognition without relational constraints, then the recognition speed is fast, but the consistency and synchronization between left and right facial parts deteriorates
Solution Approach 1:
The patent applies asymmetry principle by treating left and right expression coefficients differently based on their similarity. When similarity is high, symmetric processing (averaging) is applied; when similarity is low, asymmetric processing (maintaining original coefficients) is applied. This resolves the contradiction by adapting the processing strategy to the specific case, ensuring consistency only when appropriate while preserving natural expression variations.
Solution Approach 2:
The patent changes the parameter of expression coefficients through similarity-based processing. By calculating similarity between left and right coefficients and using it as a basis for generating target coefficients, the system dynamically adjusts the expression parameters to ensure consistency. This parameter transformation resolves the contradiction between maintaining reliability and managing complexity.
2Reliability
If similarity-based processing is applied to generate target expression coefficients, then the synchronization between left and right facial parts is improved, but the computational complexity increases
Solution Approach 1:
The patent transforms expression coefficients into similarity values and then into target coefficients through controlled parameter changes. This approach improves synchronization by ensuring left and right coefficients are related, while the parameter transformation process is designed to be computationally efficient, balancing reliability improvement with acceptable processing time.
3Reliability
If direct expression recognition is used without similarity processing, then the processing efficiency is high, but the authenticity of expression driving deteriorates
Solution Approach 1:
The patent uses asymmetric processing where the degree of coefficient adjustment depends on the similarity metric. High similarity cases receive symmetric adjustment for authenticity, while low similarity cases maintain original coefficients for efficiency. This resolves the contradiction between authenticity and processing efficiency by applying different strategies selectively.
Solution Approach 2:
The patent introduces similarity as an intermediate parameter that controls the transformation from initial to target expression coefficients. This parameter-based approach enhances authenticity by ensuring coefficient consistency while maintaining processing efficiency through a straightforward calculation pipeline, resolving the contradiction between authenticity and productivity.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
Embodiments of the present disclosure relate to an expression driving method and apparatus, a device, and a medium. The method includes: obtaining a target image, and recognizing an expression of a target object in the target image to obtain at least one expression coefficient; extracting an expression coefficient to be processed from the at least one expression coefficient, and determining an initial left expression coefficient and an initial right expression coefficient in the expression coefficient to be processed; generating a target left expression coefficient and a target right expression coefficient based on a similarity between the initial left expression coefficient and the initial right expression coefficient; and driving a virtual character to show a corresponding expression based on the target left expression coefficient and the target right expression coefficient.