Text-To-3D Scene Generation With Inpainting and Depth Diffusion Priors