Artist skills being taken out?

[Atownsend]

That's awesome, but still where is the art coming from is AI looking on the web and pulling art that someone has posted or does it create the art?

The way that I’m understanding it is that the GPT references massive datasets of text image pairs based on the prompt inputs. Like several billion. Then it generates new imagery based on the inputs and cross references. The output is unique, but the reference materials are in the public domain I’d imagine.

[Dottonedan]

I've some across two different situations of creating different art...where it has generated the art (or a portion of) with an artist signature at bottom right or left). I can only assume that it's only referenced "some of that" art and combined with it others. I've read where several artist have seen their own art being used in a large way apparently and had sued. Don't know how that ended.

[GraphicDisorder]

I've some across two different situations of creating different art...where it has generated the art (or a portion of) with an artist signature at bottom right or left). I can only assume that it's only referenced "some of that" art and combined with it others. I've read where several artist have seen their own art being used in a large way apparently and had sued. Don't know how that ended.

It will certainly eventually generate a lawsuit that sets this straight on what is going to be ok or not.

[ebscreen]

The output is unique, but the reference materials are in the public domain I’d imagine.

The output is unique, but the reference materials are searchable by Google I’d imagine.

FTFY

[Dottonedan]

Mid Journey has a tab where you can look at Google examples of a similar nature. (It doesn’t say) of those are what it’s pulling from.

[Atownsend]

The output is unique, but the reference materials are in the public domain I’d imagine.

The output is unique, but the reference materials are searchable by Google I’d imagine.

FTFY

I cant find stupid memes that I remember from just a couple years ago, so im not sure how likely that is when the dataset is so large. Maybe I just suck at search.

I dont see how its much different that how a human artist operates. Very little in this world is actually original. All of the artists I've have worked with in the past have their process. One guy I worked with would go to Walmart, pose some action figures how and then redraw them adding fonts and text styles. Theres almost always a reference image involved. Fact is most designers I come across can't or wont produce high quality photo realistic graphics. Most don't have the skillset, and so this fills a lot of the gap. There are exceptions, no doubt. Brant has a killer art dept. and I dont think this is is really competitive with that. Its still clipart vs hand drawn.

With that being said we're still putting in a lot of work, and this particular design and print is going to be much different than the original output from AI. Its not as simple as copy / paste. Its just a tool in the toolbox, you still need to be very proficient in the raster / vector programs. I don't that its taking away the artists skill as much as its allowing lesser skilled or talented people to come up a bit. We can disagree, but I think its a really cool tool thats only going to continue to advance.