Articulated 3D Head Avatar Generation using Text-to-Image Diffusion Models