Nonlinear Blend Shape Combination for Realistic Facial Animation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional blend shape animation techniques often result in unrealistic facial expressions due to interference from multiple displacements applied to common vertices, leading to overextension of features like eyebrows, which can appear physically unrealistic.
Innovation Solution
Implementing nonlinear operations, such as calculating the expected value and applying biases to blend shape displacements, in combination with linear operations, to reduce interference and produce more realistic facial expressions by averaging or weighting displacements accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple blend shapes are combined using linear operations, then the complexity of facial expressions is improved, but unrealistic deformations occur due to interference from multiple displacements applied to common vertices
Solution Approach 1:
The patent changes the operational parameters from linear to nonlinear operations when combining blend shapes. Specifically, it applies nonlinear blending functions that consider the interactions between multiple displacements, transforming how vertex positions are calculated when multiple expressions are combined. This resolves the contradiction by maintaining expression complexity while preventing unrealistic deformations through mathematically sophisticated parameter transformation.
Solution Approach 2:
The patent introduces an intermediary computational layer between the input blend shapes and the final output. This intermediary applies nonlinear operations that mediate the interaction between multiple displacements, preventing direct interference that causes unrealistic deformations. The nonlinear blending acts as a mediator that reconciles conflicting displacement vectors while preserving the intended facial expressions.
2Manufacturing precision
If vertex positions are adjusted to represent different facial expressions, then the realism of facial animations is improved, but the number of vertices required increases significantly
Solution Approach 1:
The patent merges multiple blend shapes into a single coherent deformation by applying nonlinear blending operations. Instead of treating each blend shape as an independent vertex displacement that requires separate high-resolution modeling, the nonlinear combination allows multiple expressions to be integrated into one unified vertex set, reducing the total number of vertices needed while maintaining realism.
Solution Approach 2:
The patent creates a universal vertex displacement system where a single set of vertices can represent multiple facial expressions through nonlinear blending. This multi-functional approach allows the same vertex structure to serve various expression needs, eliminating the requirement for separate high-resolution vertex sets for each expression type.
3Manufacturing precision
If nonlinear operations are applied to combine blend shapes, then the realism of facial expressions is improved, but the computational complexity increases
Solution Approach 1:
The patent segments the nonlinear blending computation into manageable components. Rather than applying a single complex nonlinear operation to all vertices simultaneously, the computation is divided into smaller steps that process different aspects of the blending separately. This segmentation reduces the immediate computational burden while still achieving realistic results through the cumulative effect of multiple simpler operations.
Data Source
AI summary
A system includes a computing device that includes a memory for storing instructions. The computing device also includes a processor configured to execute the instructions to perform a method that includes combining, in a nonlinear manner, a first set of vertex displacements that represent the difference between a first animated expression and a neutral animated expression with a second set of vertex displacements that represent the difference between a second animated expression and the neutral animated expression. The number of vertices associated with the first set of vertex displacements of the first animated expression is equivalent to the number of vertices associated with the second set of vertex displacements of the second animated expression.