3D Caricature

3D caricature generation aims to create exaggerated, yet recognizable, three-dimensional representations of human faces. Current research focuses on developing methods that allow for both automated creation and user-guided manipulation of these caricatures, often employing neural networks like multi-layer perceptrons (MLPs) and leveraging techniques such as hypernetworks and attention mechanisms to control the degree and type of exaggeration. This work addresses challenges in balancing identity preservation with stylistic exaggeration, and also explores the potential for bias and stereotype reinforcement in AI-generated caricatures, particularly within the context of large language model simulations of human behavior. The resulting advancements have implications for both artistic applications and the critical evaluation of AI-driven simulations of human characteristics.