I mean for the case of putting pictures and characters with certain poses make it more appealing and add some level of control when your machine makes an art. Granted we can describe whatever we want and the machine will produce it but that's the limit of our control. At least if we implement ControlNet, we again have some degree of control (i.e. if we our picture to come out with certain pose or form)

Ya interesting, I can see it being helpful for stock photography and such to get more exactly what you want.

Suggest adding it to the roadmap ... I see it being very useful

https://artsmart.ai/roadmap

Cheers!