Technology is not the problem, protecting creators is

[trappedslider]

Is that really viable? It's been a while since I worked with neural networks, but when I did you needed a fair amount of data to train them. Is a single artist going to be able to produce enough data?

It's an open question, and completely depends on your dataset quality, how well you caption your data, how noticeable or subtle the style is, what range the style can have, what base model you are training from, etc. The quickest way to get your answer is to start training.

Some use as little as for 4 images, as like one model that's trained using 11 million public domain images, Once you get going it's more fine tuning than anything.

 

[overgeeked]

It's not that hard to understand. Most people, especially the masses, tend to root for the underdog in most cases.

Metallica vs fans downloading music. Most people side with the fans.

Multinational billion-dollar megacorps vs artists on Deviantart. Most people side with the artists.