Computer Animation Interaction Design Driven by Neural Network Algorithm

Computer animation interaction design in traditional research has a single user experience and cannot adapt to personalized needs due to problems such as disconnection between animation and user interaction, insufficient intelligent recognition ability, and poor real-time response. This article uses a design method driven by a neural network algorithm to achieve a deep integration of animation and interaction, and improve the naturalness, fluency and real-time performance of animation. In the study, GAN (Generative Adversarial Network) is used to generate natural and smooth animation movements, and CNN (Convolutional Neural Network) is used to analyze users’ facial expressions and voices. LSTM (Long Short-Term Memory) is used to process users’ time series data; RNN (Recurrent Neural Network) is responsible for real-time analysis of users’ dynamic input and generating instant animation feedback; TCN (Temporal Convolutional Network) optimizes the system’s real-time feedback mechanism and improves the response accuracy and speed to continuous input. Compared with the FSM (Finite State Machine) and BT (Behavior Tree) methods, the maximum response time of this method is 0.174s; the DTW (Dynamic Time Warping) distance value is between 0.136 and 0.402; the maximum Euclidean distance is 0.416. Experimental results show that through the combination of these algorithms, the system can generate personalized animation feedback according to the user’s real-time input, emotional state and behavioral characteristics, significantly improving the naturalness, fluency, real-time and personalized response capabilities of the animation, and providing a more intelligent, natural and immersive interactive experience.