AI Repaints Famous Pictures (project stack=3 :( )

 Everyone seems to think you just type what you want into gen AI and get amazing output... Firstly that's not true - there's a lot of work needed to get good outputs, and secondly (and crucially to me as an artist) - you may get good looking outputs, but it's very hard to control what that output looks like - if you want A picture of a curved bridge over a pond with waterlilies, go fill your boots... If you want Monet's picture of a curved bridge over a pond with waterlilies, it gets a lot harder... And if you want the specific bridge that is in your imagination... then it can be very frustrating to say the least...

So to try and prove a point, I thought it would be fun to describe a few famous pictures to Stable Diffusion, and see what it paints... The photoshop them back into the gallery - to try to give non-artists a taste of what it feels like when you don't get what you are expecting...

So far I've done "The Haywain" (aka ""painting of a horse and hay cart fording a river. a mill behind, a dog watching, open fields behind, trees, english countryside"):

And Monet's Waterlilies (aka "impressionist pastel picture of a curved white bridge over a pond filled with water lilies, willow trees on each side"):

Stable Diffusion does a better job than I expected, especially for the waterlilies... But I suspect it was trained on the latter, so maybe it's not a fair test... Nice bit of fun though. I should do a couple more, post something snarky on LinkedIn and get back to me root ;)