Style Motion Conversion for Realistic 3D Character Animation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in creating diverse and realistic three-dimensional object movements, particularly for character animation in films and games, as motion capture alone is insufficient for achieving various styles of human movements.
Innovation Solution
A motion conversion device that extracts content and style features using a large-scale language model and VLP model, generating style motions by combining these features through a process involving content feature extraction, style feature extraction, and style motion generation, with the aid of models like AdaIN for refining the style application.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If motion capture technology is used to obtain human movements, then realistic character animation can be achieved, but it is very difficult to obtain various styles of human movements
Solution Approach 1:
The patent segments the motion data into content features (basic motion information) and style features (characteristic motion patterns). By separating these features, the system can independently process and recombine them to generate various styled animations while maintaining realistic base movements, thus resolving the contradiction between motion variety and realism.
Solution Approach 2:
The patent transforms motion data into feature space representations where style parameters can be adjusted independently. By changing style parameters (such as motion intensity, speed, or characteristic patterns) while keeping content parameters fixed, the system generates diverse movement styles from the same realistic motion capture data, achieving both versatility and reliability.
2Adaptability or versatility
If multiple models (first model, second model, third model) are used for feature extraction and style generation, then diverse and realistic style motions can be generated, but the device complexity increases
Solution Approach 1:
The patent designs a multi-model system where each model serves multiple functions: the first model extracts content features for motion preservation, the second model extracts style features for characteristic expression, and the third model synthesizes them. This universal approach allows the same model architecture to handle different aspects of motion analysis, reducing overall system complexity while maintaining diversity in output.
Solution Approach 2:
The patent introduces feature extraction models as intermediaries between raw motion capture data and the final styled animation. These intermediary models transform complex motion data into simplified feature representations that are easier to process and combine, thereby managing system complexity while enabling diverse style generation through controlled manipulation of intermediate features.
Data Source
AI summary
Disclosed is a motion conversion device based on style and a method thereof, the device may extract a content feature from content motion data including a motion of an object using a content feature extraction model, extract a style feature from style information using a style feature extraction model, and generate a style motion reflecting the content feature and the style feature using a style generation model.