When someone renders out several angles of the exact same 3d scene; exact same poses, etc.
- Develop some artistic vision and make up your mind
- Stop giving the medium a reputation for being spammy
- I feel like it’s something people tend to grow out of, so if you’re planning to make multiple views, your scene probably isn’t interesting enough to deserve it in the first place
I’m not spoiling AI stuff entirely because I like the pony voices, but if I could just spoil the AI generated art, I would be incredibly happy.
AI voice stuff is getting tagged
machine learning assisted, so it should be pretty reliable for your preferences to spoil/hide
machine learning generated. If you want to go a step further, you could also use a line in the complex filter for
machine learning assisted, -webm
because voice stuff will always be WEBM.
Another reason I put this on spoiler, other than recognizing it’s an AI art, is the fact all arts done with it looks repetitive. No matter which art, they all look the same, even if it’s a different character.
Hyperbole, and trivial to show counter-examples. I’m really tired of the not-takes thrown around in arguments about AI art, and it doesn’t help your point.
Stablediffusion, NovelAI arts kinda became laughably recognizable for me these days. Because everyone keep putting the word “masterpiece” or insert popular artist’s name here and without fail the AI generate the same looking art with the smooth shading style all the time, to the point it’s becoming homogenized.
Now this is fair to say, and constructive. Yeah, I agree, to an extent. NovelAI in particular was making a lot of very same-y pictures, especially in terms of the faces and shading. I ended up hiding AI art on Pixiv because of this very thing: overuse and under-variety. IMO it’ll be critical for people to have very high standards when it comes to AI art; the capacity for bulk output necessitates that.
I’m not sick of the art per se…it’s just that when I see some detailed, professionally rendered image on this board I immediately want to congratulate the artist–but who the hell do I praise when it says “AI ART”?! How can I ask the artist “how did you do it, how long did it take you in years to get this good?” when the answer is a very obvious “I didn’t do it, but it took me a minute.”
Well…at least it says “AI art”.
You can’t form a fair opinion when you don’t know what went into a piece:
“Wow, this game world looks amazing, it’s so lifelike!” It’s assets from Megascans…
“Wow, the inkwork is so smooth, you must have so much practice!” It’s stabilizer…
“Wow, the clouds are so beautiful, what’s your technique?” It’s a stock photo…
All useful, all have their place, all things someone might or might not value. When you learn a piece is generated by AI, you can heavily discount the importance of “detailed” or “quality rendering” and evaluate things like “what is depicted/being conveyed”.
How does it work? which library of images does it take from, what sort of transformation did it do, is it Fair Use, etc.
If many amateur and beginner artists including me, don’t know where the line between inspired by and plagiarized from is…does the AI know?
Yeah, this is one of the concerning things to me about this tech. It’s not supposed to copy any specific thing (not by the research goals, anyway - of course, people can misuse it) but how would anyone know if it does produce something too similar to a pre-existing work, by accident? It’s not feasible to be familiar with all of the hundreds of millions of images in the data sets. At least with good old-fashioned tracing and plagiarism, someone involved knew what they were doing.
It also hits really different in my mind when it’s individuals using and training the software, vs. when it’s benefiting companies selling its use as a product (eg. that portrait app that was getting press recently).