Engineering Animated Characters for Cross-Cultural Markets: AI-Assisted Parametric Modeling and Cultural Identity Adaptation

Abstract

3D animation character modeling is not only time-consuming and labor-intensive, but also has high requirements on the modeler's professional skills. For this reason, this paper proposes a parametric model regression algorithm based on a two-branch structure from the problems existing in animation character modeling. The algorithm obtains the corresponding shape parameters and pose parameters as a priori information by matching the parametric model, so that the implicit function can be effectively sampled and mapped to the point set in the reconstruction. At the same time, the shape parameters and pose parameters of RaBit are regressed separately, and multiple rounds of regression are performed on the pose parameters using the geometric features as reference information in order to realize the fine-tuning of the pose parameters. On the three datasets of Render People, THUman 2.0 and Deep Human, the PSNR values in terms of reconstruction accuracy are 30.83, 29.93 and 31.08, and in terms of rendering efficiency are 64, 48 and 73 FPS, respectively, which achieve a good balance between reconstruction accuracy and rendering efficiency. Finally, the value of animation characters in spreading and transmitting national culture and strengthening cultural identity is discussed, and a cultural identity suitable for the development of domestic animation is constructed, which provides a new direction for animation character reconstruction.