I trained SD 1.5 on old monographs from different archives online. I trained it on peoples faces, buildings, animals and some medical diagrams of the human body as well. I used LORA to train it. I also generated these with deliberate model. In order to use, you must have the LORA extension installed in Automatic111. This works really well with other models that are trained on your face for example.

https://github.com/kohya-ss/sd-webui-additional-networks