DiffEdit: Diffusion-based semantic image editing with mask guidance - 42Papers