Avatar Expression Segmentation for Faster Facial Animation Loading

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional methods for loading expression data into avatar models result in heavy workloads and inefficiencies due to the need to load all expression data, including unnecessary data, which affects the calculation and loading efficiency of animated avatars.

Innovation Solution

A method that determines the location of a human face in an image, obtains current expression data, identifies a target split expression area, and loads only the necessary expression data into that area to update the avatar's expression, using a combination of basic and mixed avatar sub-models to reduce calculation and improve efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If all expression data of the avatar model are loaded into the avatar model, then the avatar model can have complete expression data, but heavy workload is caused in expression data calculation and loading efficiency is reduced

Engineering Contradiction:
Improveexpression dataVSAvoidloading efficiency
Core Design Contradiction:
Quantity of substanceVSProductivity

Solution Approach 1:

The avatar model is divided into multiple split expression areas (e.g., eyebrow area, eye area, nose area, mouth area, cheek area). Instead of loading all expression data uniformly, the system segments the expression data according to these areas and loads only the required segments into the corresponding split expression areas, thereby improving loading efficiency while maintaining complete expressive capability.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system extracts only the necessary expression data from the complete expression data set based on the target split expression area that needs to be updated. By taking out only the required expression data corresponding to the specific expression area, the system avoids loading unnecessary data, thus reducing workload and improving loading efficiency.

Inventive Principle:
Principle #2Taking out (Extraction)

2Quantity of substance

If all expression data of the avatar model are loaded into the avatar model, then the avatar model can have complete expression data, but heavy workload is caused in expression data calculation

Engineering Contradiction:
Improveexpression dataVSAvoidcalculation workload
Core Design Contradiction:
Quantity of substanceVSLoss of energy

Solution Approach 1:

The expression data calculation is segmented by expression area. The system determines the target split expression area based on the expression change data, then performs calculation and loading operations only on that specific segment. This segmentation reduces the overall calculation workload compared to processing all expression data uniformly.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system performs partial action by calculating and loading expression data only for the necessary target split expression area rather than performing excessive action on all expression data. This partial processing approach significantly reduces calculation workload while still achieving the desired expression update effect.

Inventive Principle:
Principle #16Partial or excessive action

3Adaptability or versatility

If basic avatar sub-models are combined through non-linear combinations to generate mixed avatar sub-models, then more diverse avatar expressions can be achieved, but device complexity increases

Engineering Contradiction:
Improveavatar expression diversityVSAvoidmodel complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The avatar model is segmented into multiple basic avatar sub-models corresponding to different expression areas (eyebrow, eye, nose, mouth, cheek). Each sub-model can be independently combined through non-linear combinations to generate mixed avatar sub-models. This segmentation approach enables diverse expression combinations while managing complexity through modular organization.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system uses composite modeling by combining multiple basic avatar sub-models through non-linear combinations to create mixed avatar sub-models. This is analogous to composite materials where different components are combined to achieve new properties - here, different basic sub-models are combined to achieve diverse avatar expressions while managing complexity through structured composition.

Inventive Principle:
Principle #40Composite materials

Data Source

PatentEP3751521B1Expression animation data processing method, computer device and storage medium
Publication Date: 2025.12.03 TENCENT TECHNOLOGY (SHENZHEN) CO LTD
  • EP3751521B1 patent drawingFigure 1~2
  • EP3751521B1 patent drawingFigure 3~4
  • EP3751521B1 patent drawingFigure 5~6

AI summary

The present application relates to an expression animation data processing method, a computer readable storage medium and a computer device, said method comprising: determining the position of a face in an image, and acquiring a virtual image model; acquiring current expression data according to the position of the face in the image and a three-dimensional face model; acquiring expression change data from the current expression data; determining a target segmented expression region matching the expression change data, the target segmented expression region being selected from segmented expression regions corresponding to the virtual image model; acquiring target basic virtual image data matching the target segmented expression region, and combining the target base virtual image data according to the expression change data, so as to generate loading expression data; and loading the loading expression data into the target segmented expression region, so as to update the expression of the virtual animation image corresponding to the virtual image model. The solution provided in the present application can realize the reduction of the computation amount of emotional data, thereby improving the emotional data processing efficiency.