AI Digital Anime Style Generation Algorithm Based on Adversarial Generative Network
Xiaojun Tan · 2024
Digital computer technology is changing the development of animation image processing and significantly improving the efficiency of animation creation. Among them, traditional artificial animation faces problems such as low efficiency and poor quality in image drawing. Therefore, a research proposes a digital animation style generation model based on artificial intelligence. First, the animation style generation model is constructed based on the generative adversarial network. Considering the gradient vanishing problem, the residual network is introduced to perfect the model, and the attention mechanism is introduced to correct the image deviation problem. In the two scenarios of contour feature extraction and comic style transfer, the image loss of the research model is 0.012 and 0.038 respectively, which is better than similar models. In addition, in the comparison of animation style conversion quality, the research model handles the details of animation images better, and its peak signal-to-noise ratio is 23.05db, which is better than similar models. It can be seen that the research technology is superior to similar technologies in the field of animation creation and has good application effects. The research content will provide a technical reference for intelligent animation creation.