In my company’s presentation template, they have an AI-generated image of the inside of a cathedral, with one of our products placed on the ground and then a component of that product in much larger size is floating in the air. And then there’s a woman in spiritual clothing looking up to that component, like it’s some sort of god.
I do have many questions, but the main one is: What the fuck?
It was added in post, I’m pretty sure. The product itself, it might’ve been able to generate, but wouldn’t have nailed all the details. And the component, there’s just no way it would have enough training data, nor would training data contain the name of that component.
Don’t worry the other apps are still full of them, and my boss still uses them for every department meeting.
The OP didn’t say lemmy doesn’t have recent AI-generated images.
The implication appears to be that OP stopped noticing them and therefore they must be impossible to tell from the real thing anymore.
In my company’s presentation template, they have an AI-generated image of the inside of a cathedral, with one of our products placed on the ground and then a component of that product in much larger size is floating in the air. And then there’s a woman in spiritual clothing looking up to that component, like it’s some sort of god.
I do have many questions, but the main one is: What the fuck?
NGL that sounds pretty hilarious, I’d like to see that.
Seconded
Thirded
EDIT: I don’t know the Lemmy rules actually - maybe seconding a motion is enough to set it into action
The second normally brings it to a vote. So … all those in favor?
… Aye.
Dox! Dox! Dox!
MOX! MOX! MOX!
…sorry, I was watching AEW.
The upvotes have it!
That’s pretty fuckin funny man, are you sure the art boys aren’t self aware?
Gotta ask, does it look like your product/ the component were part of the prompt, or were they added in post?
It was added in post, I’m pretty sure. The product itself, it might’ve been able to generate, but wouldn’t have nailed all the details. And the component, there’s just no way it would have enough training data, nor would training data contain the name of that component.
I see. I think that makes it better, honestly