- Joined
- Aug 28, 2019
Hopium. Also how about you people archive your shit.
JT 🔜 FWA/AC on Twitter: "Support artists in our fandom that actually…
archived 14 Apr 2023 20:02:12 UTC
AI Art Seething General
- Thread starter The Thinking Abacus
- Start date
>It can only replicate-
As long as it can be described people will probably narrow it down to an algorithm. The only thing AI can't do is something not well described and in the realm of subjectivity, but that is something incapable to replicate by humans either. If someone can distinguish between the styles of different artists, AI, with just machine learning a few years ago can probably figure it out as well. How long do you think it will be that AI's will be able to succinctly describe a piece of medium verbally and you can put that back into a prompt? That is more complicated but it doesn't sound outlandish anymore considering we've passed several levels of absurdity with AI.
Really the only thing that will prevent machine's reaching complete human capabilities, is that it might turn out that the cost of development of making AI completely human isn't worth the benefit of making a completely human-like AI. I mean are companies going to invest in a machine that contemplates resisting its owners, much less a machine that is capable of unionizing?
DJ10
kiwifarms.net
- Joined
- Mar 31, 2021
That's a shame.
I did actually but it didn't show the responses I was talking about, I just figured generic smug furfag tweet #274856849 wasn't worth the effort, you are correct though my bad.
- Joined
- Aug 28, 2019
It seems widespread in this thread, not just you.
- Joined
- Jun 5, 2020
- Joined
- Aug 17, 2020
AIs have convinced me that neither does the majority of the human race have a 'soul' or whatever.
- Joined
- Feb 13, 2022
I kinda wonder how well AI will get in terms of scene generating or panel handling in regards to the context of comics. Like take for instance stonetoss comics, if you fed an AI all of his comics and churned them out, they'd probably look like incomprehensible splotches with no cohesion since ST very meticulously (no better word to describe it sry) positions and draws his comics with intent and a vision in his mind over how the joke will go. The AI doesn't know how Stonetoss comics will play out, at best it just as a general idea over how his artstyle is like and not how a standard 'toss comic is structured. And this goes for other comics on the web as well with more or less simplified artstyles.
Of course if those artists were smart, they'd use AI to create less workload and fix up any mistakes the output generated and fix the positioning, joke structure, word bubbles etc. etc. But for right now it seems like those making comics will either have to wait until whatever image generation program gets better or train their own AI program to draw for them, however long that would take.
Of course if those artists were smart, they'd use AI to create less workload and fix up any mistakes the output generated and fix the positioning, joke structure, word bubbles etc. etc. But for right now it seems like those making comics will either have to wait until whatever image generation program gets better or train their own AI program to draw for them, however long that would take.
Hair and Nails
kiwifarms.net
- Joined
- Mar 25, 2021
To me, this seems like a solvable problem. You don’t train blindly on comics, but panels, and then you do the layout yourself. Stonetoss *panels* are quite formulaic, and the panel layouts are pretty standard, too. I’m sure you could go one level higher and have the AI learn how to do a panel layout, gutter size, speech bubbles, but it’s easier to do that yourself. Most AI comics seem to have gone with this approach so far.
- Joined
- Dec 28, 2014
That actually does present legal issues especially if it's explicitly ripping off a style, and even more so if it's creating literal fakes and otherwise acting as a market substitute, not necessarily as a direct copyright claim but maybe as a right of publicity claim of some sort (like the Tom Waits case where he sued for ripping off his voice with an imitator for a Dorito's ad he'd refused to do for them).
You'd really have to have a distinctive almost trademarkable style to make such a claim though. Most furfag art and animu shit is so samey nobody really owns it, and it's mainly these seething whiners who are complaining. If you ripped off something more specific with commercial value, such as ripping off the Sanrio style, they could probably figure out something to go after you for, although the contours of the legal doctrines will probably take years to develop.
So if according to this mongoloid, AI art is literal shit, then what does it say if his own art is indistinguishable from literal shit?
Very nice self-own though.
Last edited:
- Joined
- Jun 5, 2020
So somebody redrew that picture I posted earlier from Twitter.
Imo it looks even more dog shit than the original AI version.
Imo it looks even more dog shit than the original AI version.
Mr.Miyagi
kiwifarms.net
- Joined
- Jul 5, 2022
Here are the two images side by side, for lazy people like me.
Having multiple characters (without them turning into blobs) is something that AI (or at least my average attempt) has some trouble with, so I think that addition is a point in favor, considering current limitations. I think the AI's has a better amount of detail (on things like clothes and the food) and use of light and shadow, but I'm not an artist, so I know fuck-all. Copying the AI's garbled "Happy" text is a cute touch.
- Joined
- Jun 5, 2020
I just personally find the “redraw” version to look more sloppy and less detailed than the AI generated version, the food is the biggest difference I first spotted as the AI has not just the noodles but there’s more to it than the vaguely slopped on shit the human™ artist painted. I will agree adding more characters in the background definitely helps make the environment more alive and fitting for the original title from the original AI artist.
- Joined
- Jun 24, 2019
Was the one on the left supposed to be the wolf from Puss in Boots, or am I just confused? The one on the right (the human drawn one) looks nothing like the movie character.
- Joined
- Feb 13, 2022
AI one looks a lot more striking, contrasted, clear and stunning. The one of the left looks more washed out and blotchy.
- Joined
- Jun 5, 2020
I believe it’s just my bias opinion for liking stark contrasting art more than smooth textureless art. The AI art could easily use some touch ups on photoshop without much artistic skill to give a bit more emphasis on the art itself and fix some errors such as the humanoid fingers and text.
- Joined
- Feb 13, 2022
Am retarded and just realized I wrote "left" instead of "right" for the human-made art. Basically I'm agreeing with you is what I'm saying.
- Joined
- Jun 5, 2020
Ngl this gave me a chuckle.
All good fam, I understood what you meant.
FBI-chan
kiwifarms.net
- Joined
- Sep 12, 2019
Your art reflects who you are, and at the time the artist was upset and coping hard.