I can open my comfyUI dashboard and generate an image that contains other images. It’s not that hard. OP mentions they’re running local models. I am inclined to believe them.