We propose a novel combination of local as well as global conditioning modules with which we finetune Stable Diffusion on the ScanNet++ dataset.