The Design and Application of Anime Game Character Modeling Using Long Short-Term Memory Network Algorithm

Abstract

The goal of this study is to utilize the Long Short-Term Memory (LSTM) algorithm to design a model for anime game character modeling, aiming to improve the effectiveness and efficiency of anime character design to meet the industry's demand for more precise and diversified designs. This study introduces the LSTM algorithm and its application in the field of image generation. It proposes the integration of multi-attention mechanisms with Bidirectional Long Short-Term Memory (BiLSTM) to enhance the model's ability to capture image details and diversity. Subsequently, the study constructs an anime game character modeling design model based on the fusion of multi-attention mechanisms and BiLSTM algorithm, followed by experimental evaluations. The experimental results show that the proposed model achieves prediction accuracy and F1 scores of 96.91% and 91.79%, respectively, in anime character design, improving accuracy by at least 3.08% compared to other models. Additionally, the Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) of the proposed model algorithm decrease errors by 6.52% to 10.7%. Therefore, the model algorithm presented in this study can provide valuable application in improving the effectiveness and efficiency of anime character design for the development of the anime industry.