General Vtuber Discussion (V1)

[RestlessRain]

In an effort to bridge the gap between subjects, I wonder if the possibility to utilize AI generated art as the models for future vtubers will become the next evolution of the medium. I don't know how the tech works when it comes to creating a vtuber model that makes expressions and what not. But I'd imagine a rigger would still need to be hired to set up all that to get it working.

It's hard to imagine how far in the future something like this becomes feasible, but it's not happening any time soon. Rigging and animation is an extremely complex, time-consuming and high-skill area of art. A high-quality vtuber model, like you'd see in a company, has over a thousand different moving layers creating credible animated movement to track with the acting that the vtuber does. Creating realistic movement in computer-generated animation is still the realm of artists, rather than AI.

 

[Tamavitch]

In an effort to bridge the gap between subjects, I wonder if the possibility to utilize AI generated art as the models for future vtubers will become the next evolution of the medium. I don't know how the tech works when it comes to creating a vtuber model that makes expressions and what not. But I'd imagine a rigger would still need to be hired to set up all that to get it working.

AI didn't generate anything, though. It pastes together existing art. So what you propose (if we want an accurate comparison) is the AI pulls Kiwawa's hair and left eye and Pippa's right eye and face oval and CTRL-P it onto Tenma and then adds some 2view's angel wings to "create" a "new" vtuber after using blur and clone stamp to ease the seamlines. So your new chuuba gets sued by at least 4 people. Which is one good reason for Holos not to encourage it in their fanart tags.
Another way to look at it is your new character was made in the World of Warcraft character creation screen. That's what the AI art is. Maybe you can get just the character you want but it's still made of Blizzard art team's assets, nothing is new. The algorithm just uses a much bigger pool of assets, consisting of millions of finished pieces.

 

[BioBreak]

I just like that we can all just use AI art as just another, albeit glorious, tool for shit-posting.

Lightnovel title: "My Pipkin Rabbit Turned Out to be a Lovecraftian Horror From Outer Space!"

 

[El Rrata]

It is coming.

 

[agility_]

Late, but, what did I miss with Nina now?

 

[Random_Cultist]

AI didn't generate anything, though. It pastes together existing art. So what you propose (if we want an accurate comparison) is the AI pulls Kiwawa's hair and left eye and Pippa's right eye and face oval and CTRL-P it onto Tenma and then adds some 2view's angel wings to "create" a "new" vtuber after using blur and clone stamp to ease the seamlines.

Oh dear God, it's Moetron: VTuber Edition.

 

[Fucking YTs]

It is coming.

Damn it. I’m going to miss thIs disaster live.

 

[Gravedigger]

This is all neither here nor there. Until the AI can make furry fetish porn of your OC fucking a Navi, seething Predditor/Deviantarts will be able to supplement their disability checks for BPD with those sweet $40 commissions. Right now all the AI does is questionably use a blend tool to past together furry porn that somebody already drew.
This is why it can't do hands, btw, and that will never and can never be fixed. The whole thing is a rain puddle people are pretending is an ocean and the real revelation is just (as someone mentioned) showing arrogant twats exactly how little the plebian really care about or notice details. Something any actual artist has known all along. It's the main thing we pompously bitch about. One person in 50 even at a gallery show knows an objectively good from bad piece (and that there is indeed an objective good and bad). That people think this AI crap is even kind of passable just demonstrates this fact. It is all laughably bad and freakish but that goes over a lot of people's heads. I liked the Star Wars characters on trail cams iteration better, which is the exact same tech.
Now at some point in the future, if society survives, there might be a system that can actually pull from some database and create pieces from scratch using knowledge of anatomy and algorithms about how cloth folds and so on and make soulless but maybe realistic art. Some far-advanced extension of CG animation, maybe. But vtuber fanart collages being posted right now aren't even close. What it is doing though is make retards trying to sell jars of piss for $75,000 have at least a slightly harder time justifying it. BUT I THOUGHT ANYTHING CAN BE ART??

I'm gonna post this reddit comment that seemed rather understandable about how AI generates pictures that are not just bits of preexisting pictures pasted together .

 

[Clown Penis]

I guess it's not a Pippa stream if Pippa fails to find baking soda for like 20 minutes, right in front of her.

 

[Tamavitch]

I'm gonna post this reddit comment that seemed rather understandable about how AI generates pictures that are not just bits of preexisting pictures pasted together .

This is objectively horseshit and as always Predditors are clueless. It IS in fact just preexisting pictures (badly) pasted together. Every eye, every finger, every strand of hair, exists elsewhere already. It's not some big leap forward.

Oh dear God, it's Moetron: VTuber Edition.

Why did you remind me that exists. It's exactly what they remind me of.

Well Pippa has now decided her parents giving her Mt. Dew instead of water for months was too restrictive and now that she's an adult, she won't be so limited :_;

 

[GOD'S STRONGEST BUILDERBEAR]

I wonder when people will start tracing artwork to post that lineart on twitter as ai art in black and white.

 

[Punished Anime Discusser]

Oh dear God, it's Moetron: VTuber Edition.

My first thought was the NHK Ni Youkoso scene where Sato and Yamazaki are trying to think up a protagonist for their VN, and they just keep adding chuuni adjectives for two minutes until they end up with Ollie, but she has autism and her parents died in a car crash while she was sick in the hospital.

 

[Scoots]

Kronii has started to show some of her powerlevel lately, which I can respect.
Guess the "controversy" broke some limiters.

Idk she was always that unhinged since the beginning for me

Oh cool you made it now the Sakana react should get a lot more use

 

[DaddyChill]

Could be the finale for Bowson. Two bosses left. I know people really dislike the Elden Ring fest with all the girls, but Ame played the game unlike any of the others. Made it enjoyable to watch and she seemed to really enjoy it.

 

[Tightpinch]

Ethyria 1st anniversary merch

 

[Hff201]

To expand on the philosophical pondering, there's something I don't think's been touched on yet - AI being used as a tool by artists. If someone generates a piece, then manually cleans up the more uncanny parts, is that AI work or should the human work be acknowledged? What about if they redo the colouring, or filter it? What if they trace the generated outline as the basis for a new piece?

AI's a distinct gimmick right now but eventually it'll be harder to draw the line. I could see it evolving into an integrity argument (think how autotune has been a contentious topic in music since its inception), but moralising it just seems like it'd be awkward long-term, and a lot of artists criticising it now will have it in their toolbox down the line.

 

[Mehk]

Late, but, what did I miss with Nina now?

View attachment 1809

Usual clickbait shit, at the beginning before they debuted Nina was trying to do a 2 calls a week thing but the other 3 would tell her they were busy. Seems she found out that they would have their own calls and not include her and thought of Nina as a "normie" so Nina felt left out(?) so decided she would do her own thing. Turns out after a while they figured out they were making Nina sad and they started including her in the talks, reason I think Millie? posted about having a call with all of Ethyria.(This all stems from the Alban collab from yesterday where she talked about it)\

Timestamp is 47:50

 

[Gravedigger]

This is objectively horseshit and as always Predditors are clueless. It IS in fact just preexisting pictures (badly) pasted together. Every eye, every finger, every strand of hair, exists elsewhere already. It's not some big leap forward.

What he is decribing is a diffusion model, which are objectivly not horseshit. I'm not here to say how big of a leap this is or isn't but the comment's discription is understandable enough. I've also spend a whole 2 minutes on google and could link you hard math on how this works, but this content currently exeeds my brain capacity, but I'm sure spending 2 minutes yourself on google, you can see that diffusion models exist and work.

 

[Aka Split]

To expand on the philosophical pondering, there's something I don't think's been touched on yet - AI being used as a tool by artists. If someone generates a piece, then manually cleans up the more uncanny parts, is that AI work or should the human work be acknowledged? What about if they redo the colouring, or filter it? What if they trace the generated outline as the basis for a new piece?

This was almost the exact same thought I was just having. I was thinking more along the lines of commissions, though. Say someone creates an AI image and they really like it, but it has some weird janky stuff in it (as they all do). Will there be artists willing and able to take correcting those errors as a commission? Or alternately for the bigger name artists; ‘can you take this AI image and re-draw it in your own style?’

Someone in Reddit was posting this (non-nude but still risqué) AI generated Mumei as an example of something that is decent, but has clear issues when you take a closer look:

Spoiler

Will there be a market for people who want to commission, ‘hey, can you fix her hand, the ahoge coming out of her feather, etc?’

But from there, then you also have the question of ownership that you brought up.

 

[BioBreak]

To expand on the philosophical pondering, there's something I don't think's been touched on yet - AI being used as a tool by artists. If someone generates a piece, then manually cleans up the more uncanny parts, is that AI work or should the human work be acknowledged? What about if they redo the colouring, or filter it? What if they trace the generated outline as the basis for a new piece?

AI's a distinct gimmick right now but eventually it'll be harder to draw the line. I could see it evolving into an integrity argument (think how autotune has been a contentious topic in music since its inception), but moralising it just seems like it'd be awkward long-term, and a lot of artists criticising it now will have it in their toolbox down the line.

Anyone can use the AI to generate some abstract background or element they can throw/include as either a major or minor element of a piece. Like throw-away wall artwork in the background while they're drawing the focal character/elements in the foreground. It's just gonna be another ubiquitous time-saving tool in digital art in the same vein as a clone-stamp and smart select on Photoshop.

Or foregoing buying stock photos to incorporate as background or element into your piece and just going ahead and generating some of your own for use - like a post-apocalyptic cityscape that you can iterate repeatedly until you can get a certain look that you can then draw your main character standing over in the foreground.