Digital Character Blending With Layered Texture and Expression Control

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for generating computer avatars or digital characters compromise on quality by using linear combinations for geometry and texture blending, leading to loss of fine details and averaging of facial features, and lack dynamic expressions and demographic estimation.

Innovation Solution

A method and system for creating a texture model involving spatially varying features, with manual, semi-automatic, or automatic identification of feature locators, layer separation, and blending of texture maps, along with regional and global blending techniques to preserve facial details and allow dynamic expressions and demographic adjustments.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If linear combinations of multiple head images are used for blending, then the generation process becomes simpler and more efficient, but fine details are lost and textures become blurred

Engineering Contradiction:
Improveblending efficiencyVSAvoidtexture detail quality
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The patent segments the texture map into multiple layers based on feature locators (eyes, nose, mouth, etc.). Each layer contains specific facial features at different positions and scales. This segmentation allows selective blending of corresponding layers from multiple head images, preserving fine details in each feature region while maintaining blending efficiency.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different blending operations to different layers locally. High-frequency detail layers (containing fine features) are blended using methods that preserve edges and details, while low-frequency base layers are blended using standard linear combinations. This local quality approach ensures that fine details are preserved in critical regions without sacrificing overall blending efficiency.

Inventive Principle:
Principle #3Local quality

2Device complexity

If linear combinations are used for geometry blending, then the process is simpler, but the resulting geometry becomes smoothed out and loses facial特征

Engineering Contradiction:
Improveblending process complexityVSAvoidfacial geometry accuracy
Core Design Contradiction:
Device complexityVSShape

Solution Approach 1:

The patent segments the geometry into multiple layers corresponding to different facial regions (base geometry, eye region, nose region, mouth region, etc.). Each geometric layer can be blended independently with appropriate weighting, allowing preservation of distinctive facial features while maintaining process simplicity.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces dynamic expression layers that can be independently controlled and blended. These dynamic layers capture facial expressions and can be adjusted to create realistic animated characters while preserving the underlying static geometric structure. This dynamic approach maintains simplicity while achieving accurate facial geometry.

Inventive Principle:
Principle #15Dynamics

3Adaptability or versatility

If more faces are used for blending, then demographic diversity is improved, but the result converges to an average face model

Engineering Contradiction:
Improvedemographic representationVSAvoidindividual facial characteristics
Core Design Contradiction:
Adaptability or versatilityVSLoss of information

Solution Approach 1:

The patent segments facial features into distinct layers (base layer, eye layer, nose layer, mouth layer, etc.), allowing selective blending of features from different demographic sources. This enables creation of diverse characters that maintain individual characteristics rather than converging to an average, as each feature layer can be independently controlled to preserve distinctive traits.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies local quality control to different facial regions, allowing different blending strategies for different features. Critical identity-defining features (eye shape, nose structure) can be preserved from specific demographic sources while other regions are blended for diversity, preventing convergence to an average face while maintaining demographic versatility.

Inventive Principle:
Principle #3Local quality

4Device complexity

If static face blending is used, then the process is simpler, but dynamic expressions cannot be created

Engineering Contradiction:
Improveexpression system complexityVSAvoiddynamic expression capability
Core Design Contradiction:
Device complexityVSAdaptability or versatility

Solution Approach 1:

The patent segments the facial model into static base geometry layers and dynamic expression layers. The static layers contain the fundamental facial structure, while dynamic layers capture expression variations. This segmentation allows simple static blending for base geometry while enabling complex dynamic expressions through independent control of expression layers.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces dynamic expression layers that can be independently animated and blended. These layers contain expression-specific geometric data that can be activated and adjusted to create realistic facial expressions. The dynamic nature of these layers enables versatile expression capability while the modular structure keeps the overall system complexity manageable.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3871194B1Digital character blending and generation system and method
Publication Date: 2026.03.11 SOUL MACHINES LTD
  • EP3871194B1 patent drawingFigure 1
  • EP3871194B1 patent drawingFigure 2
  • EP3871194B1 patent drawingFigure 3

AI summary

A method for creating a model of a virtual object or digital entity is described, the method comprising receiving a plurality of basic shapes for a plurality of models; receiving a plurality of specified modification variables specifying a modification to be made to the basic shapes; and applying the specified modification(s) to the plurality of basic shapes to generate a plurality of modified basic shapes for at least one model.