SIC3D: New AI Model Generates Styled 3D Objects From Text and Images

Researchers introduce SIC3D, a two-stage pipeline that combines text prompts with reference style images to generate 3D objects using Gaussian Splatting technology. The new Variational Stylized Score Distillation loss improves texture quality and geometric accuracy, addressing previous limitations in controllability and texture clarity in text-to-3D generation.