It is a bit of cheating, and if you are a commercial photographer maybe shooting a house for a customer, it may be ok. On the other hand it is a step in the right direction because you are cataloging clouds from the region rather than pulling some cloud formation out of a library, which may never happen in the region. Next step would be if you were an expert meteorologist, you might make sure other features lined up. Next you need to examine shadows especially if the % if the sky clouded is significant. If you don't likely the image will start looking un-natural. Next you need to examine reflections, specular and diffuse, and make sure those match the clouds that you are adding, and modify all of this from what you actually took with a cloudless sky. You should also consider the contrast, both at a local as well as global level because the clouds will modify this in both local and global ways. After spending a week or two on your image with your PhD in meteorology, optics, cloud science, etc. you might end up with something realistic looking. Or you could just wait 5-10 years and let AI take care of it for you. On the other hand you could just take shots as they occurred and get natural looking images, perhaps lacking features you would prefer but maybe you could use creative approaches to mitigating the lack of these features (like avoiding having too much sky if clouds are the feature you desire).