3D Head Model Expression Transfer via Deformation Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The reliance on manual face sculpting for generating blend shapes in 3D animation results in low automation and excessive human resource consumption due to the variability in facial features across different three-dimensional head models.
Innovation Solution
A method for automatically transferring a model expression from a source head model to a target head model by performing deformation matching, determining a deformation parameter, and adjusting the target head model to achieve the desired expression, utilizing computer vision technologies and pre-trained models like Swin-Transformer, vision transformer, and Masked Autoencoder.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If manual face sculpting is used to generate blend shapes for different three-dimensional head models, then model expression construction can be achieved, but automation is low and human resource consumption is excessive
Solution Approach 1:
The patent transfers model expressions from a source head model to a target head model by copying deformation parameters and blend shape data. The expression transfer module copies the second model expression from the source head model to the target head model, eliminating the need for manual sculpting for each new model while maintaining expression quality across different head geometries
Solution Approach 2:
The patent determines deformation parameters by comparing geometric features between source and target head models. The deformation parameter determination module calculates transformation parameters based on geometric correspondence, enabling automatic adaptation of expressions to different head models through parameter transformation rather than manual reconstruction
2Manufacturing precision
If manual face sculpting is performed for each different three-dimensional head model, then accurate model expressions can be obtained, but the process is time-consuming and inefficient
Solution Approach 1:
The patent performs preliminary deformation matching between the source and target head models before expression transfer. The deformation matching module pre-calculates geometric correspondences and transformation relationships, which are then used to efficiently transfer expressions without requiring time-consuming manual sculpting for each model
Solution Approach 2:
The patent copies verified expression data from the source head model to the target head model through automated transformation. This copying process maintains expression accuracy by preserving the deformation relationships while adapting them to the target model's geometry, eliminating repetitive manual work
3Ease of manufacture
If blend shapes are generated through manual face sculpting, then model expressions can be constructed, but the process requires excessive human resources and labor
Solution Approach 1:
The patent enables the system to automatically generate and transfer model expressions without human intervention in the sculpting process. The expression transfer system performs deformation matching, determines deformation parameters, and transfers expressions automatically, making the system self-sufficient and eliminating the need for skilled artists to manually sculpt each model
Solution Approach 2:
The patent replaces the manual mechanical process of face sculpting with an automated computational system. Instead of artists manually manipulating 3D models, the system uses deformation matching algorithms and parameter transformation to automatically generate and transfer expressions, substituting human manual labor with automated processing
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
This application discloses a model processing method, apparatus, and device, a storage medium, and a computer program product. The method includes: obtaining a source head model in response to an obtained expression instruction; performing deformation-related adjustment matching on the source head model to obtain a destination head model according to a pre-defined model feature of a target head model, and determining a deformation parameter according to the destination head model and a deformation relationship from a first model expression to a second model expression in the source head model, the first model expression in the source head model being determined according to the facial expression-neutral feature data of the source head model, and the second model expression in the source head model being determined according to facial expression feature data indicated by the expression instruction in the plurality of pieces of facial expression feature data; and performing expression adjustment on the target head model according to the deformation parameter, to obtain the target head model having the second model expression. The second model expression can be automatically transferred between models, and the second model expression can be automatically generated for the target head model.